可追溯论文图样本库

最早核实视觉日期：2022-03-18；arXiv v1：2022-03-18；CVPR 2022 fixed official PDF

已查看固定PDF及目标图；以最早已核实匹配版本日期作时间代理，不声称实际作图日期。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Lei Zhang；137335；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[75, 70, 523, 191]，PDF点、左上角原点。

Figure 2. Pipeline of DW. The left part shows the overall detection model which consists of backbone, FPN and detection head. The
outputs from the classification branch (H × W × C) and the centerness branch (H × W × 1) are multiplied as the final cls score. The
box refinement module utilizes the four predicted boundary points (H × W × 8) to adjust the coarse prediction (H × W × 4) to finer
locations.The right part shows the weighting process. Given an object, a candidate anchor bag is first constructed by selecting the anchor
points near the object center. Each anchor will then be assigned a pos weight and neg weight from different aspects.

最早核实视觉日期：2022-03-18；arXiv v1：2022-03-18；CVPR 2022 fixed official PDF

已查看固定PDF及目标图；以最早已核实匹配版本日期作时间代理，不声称实际作图日期。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Lei Zhang；137335；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[314, 215, 540, 357]，PDF点、左上角原点。

Figure 1. An illustration of the difference between the proposed
DWmethodandexistinglabelassignmentmethods,e.g.,GFL[21]
and VFL [43]. For ambiguous anchors B, C and D, GFL and VFL
willassignnearlythesameposandnegweightsto{B,D}and{C,
D}, respectively. In contrast, our DW assigns a distinct (pos, neg)
pair for each anchor.

最早核实视觉日期：2022-05-03；arXiv v1：2022-05-03；CVPR 2022 fixed official PDF

已读固定PDF图注、正文及作者单位。v1与会议F完全相同；M图参考文献编号由27/49改成29/51，保留历史裁剪证据，会议M采用官方Bibtex公开月份，不能声称v1已含更新的编号。完整导出已目检。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Yali Wang；17994；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[98, 71, 497, 238]，PDF点、左上角原点。

Figure 2. An overview of Target-perceived Dual-branch Distillation framework. To simplify the description, ROI refers to the operation
to get proposal features of each image. First, a source domain image is transferred into target-like domain. All of the images from three
domains are fed to a shared proposal extractor to get proposals and proposal features. Then, the proposal features of source and target-like
images are used to train corresponding branches with supervision of ground truth. Moreover, we feed the proposal features of a real target
domain image into both branches, for learning object knowledge from both source and target-like domains. As the images from target
domain are not annotated, the model is optimized by self-distillation.

最早核实视觉日期：2022-06；arXiv v1：2022-05-03；CVPR 2022 fixed official PDF

已读固定PDF图注、正文及作者单位。v1与会议F完全相同；M图参考文献编号由27/49改成29/51，保留历史裁剪证据，会议M采用官方Bibtex公开月份，不能声称v1已含更新的编号。完整导出已目检。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Yali Wang；17994；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[309, 275, 546, 383]，PDF点、左上角原点。

Figure 1. Two typical examples of detection results on the adverse
weather conditions adaptation experiments with different methods.
Semi-supervised method UBT [27] lacks awareness of objects in
the fog. Adversarial based GPA [49] attempts to exploit the objects
in the fog but gives some wrong predictions, such as the motorcy-
cle in the first row and the person in the second row. Our methods
can predict the boxes and categories more accurately.

最早核实视觉日期：2022-04-15；arXiv v1：2022-04-15；CVPR 2022 fixed official PDF

已阅读会议版图注及正文并逐张查看导出。会议图与对应arXiv历史裁剪像素一致（MAE=0）。arXiv v1即为最早核实匹配版本。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Lei Zhang；137335；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[49, 70, 552, 237]，PDF点、左上角原点。

Figure 2. The pipeline of our proposed DenSe Learning (DSL) based SSOD method. The training data contain both labeled and unlabeled
images. During each training iteration, a teacher model is employed to produce pseudo-labels for weakly augmented unlabeled images.
In anchor-free based detectors like FCOS [44], each spatial location of the dense predictions will be assigned with one label, and the
model performance is sensitive to noisy pseudo-labels. To alleviate this problem, an Adaptive Filtering strategy is proposed to split the
pseudo-labels into three types, including background, foreground and ignorable regions. Moreover, there exist some false positive cases,
which have higher scores but are obviously wrong predictions. Thus, a MetaNet is proposed to refine these cases. To improve the model
generalization capability, unlabeled images are patch-shuffled and consistency regularizations are applied on these images with different
scales. For improving the stability and quality of pseudo-labels, the teacher model is updated by the student models via aggregation, called
Aggregated Teacher. After obtaining the fine-grained pixel-wise pseudo-labels, the detector can be optimized by the final loss, which is the
sum of L ,L and L .
s u scale

最早核实视觉日期：2022-04-15；arXiv v1：2022-04-15；CVPR 2022 fixed official PDF

已阅读会议版图注及正文并逐张查看导出。会议图与对应arXiv历史裁剪像素一致（MAE=0）。arXiv v1即为最早核实匹配版本。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Lei Zhang；137335；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[84, 78, 239, 193]，PDF点、左上角原点。

Figure 3. The distributions of TP+, TP- and BG when using 10%
labeled data on COCO. ‘TP+’ means that the estimated instance
has the same class ID as the ground-truth (GT) and the IOU of
BBox is above 0.5. ‘TP-’ means that the estimated instance has
the same class ID as GT but the IOU of BBox is below 0.5. ‘BG’
means that the estimated instance belongs to the background or
has wrong class ID.

最早核实视觉日期：2021-11-25；arXiv v1：2021-02-24；CVPR 2022 fixed official PDF

已查看v1（2021-02-24）、v2（02-25）、v3（11-25）及v4（2022-03-31）。v1/v2为蓝橙简图，无区域加权；v3首次出现与会议版匹配的绿橙架构、定位头、区域权重及图例。采用v3日期，未将重绘图追溯到v1。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Ming-Ming Cheng；85763；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[49, 70, 548, 212]，PDF点、左上角原点。

Figure 3. Illustration of localization distillation (LD) for an edge e ∈ B = {t, b, l, r}. Only the localization branch is visualized here.
S (·, τ ) is the generalized SoftMax function with temperature τ . For a given detector, we first switch the bounding box representation
to probability distribution. Then, we determine where to distill via region weighting on the main distillation region and the valuable
localization region. Finally, we calculate the LD loss between two probability distributions predicted by the teacher and the student.

最早核实视觉日期：2021-02-24；arXiv v1：2021-02-24；CVPR 2022 fixed official PDF

已对照arXiv v1：大象与冲浪板两图及红色虚线歧义标记一致，会议版仅等比调整显示尺寸与排版。图注文本抽取含PDF隐藏的(a)字符，依据可见图注转录并保留原始图注PDF。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Ming-Ming Cheng；85763；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[320, 226, 536, 321]，PDF点、左上角原点。

Figure 1. Bottom edge for “elephant” and right edge for “surfboard” are ambiguous.

最早核实视觉日期：2021-11-29；arXiv v1：2021-06-22；CVPR 2022 fixed official PDF

已阅读固定版本图注和正文；与arXiv历史对应视觉比较，具体差异记入逐图核查记录。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Laura Leal-Taixé；26825；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[53, 75, 407, 172]，PDF点、左上角原点。

Figure 2. Overview of our method.
Wefirsttrainthenetworkinasemi-
supervised manner. During active
learning, for each image, we use
the network to compute the acqui-
sition function and based on it de-
cide if we actively label the image,
if we pseudo-label it, or if we only
use it as part of the unlabeled data
for the next training cycle.

最早核实视觉日期：2021-11-29；arXiv v1：2021-06-22；CVPR 2022 fixed official PDF

已阅读固定版本图注和正文；与arXiv历史对应视觉比较，具体差异记入逐图核查记录。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Laura Leal-Taixé；26825；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[309, 216, 547, 342]，PDF点、左上角原点。

Figure 1. An object from a low-performing class (Pottedplant),
original image in blue, augmented version in orange. a) Because
of its low-entropy, uncertainty-based AL methods do not label the
image. b) Because of its high confidence, pseudo-label SSL meth-
ods wrongly pseudo-label the image and thus harm the training. c)
Because of its high-inconsistency, consistency-based SSL meth-
ods cannot learn from it. d) Our method selects the image for
labeling and prevents it from getting pseudo-labeled.

最早核实视觉日期：2022-03-15；arXiv v1：2022-03-15；CVPR 2022 fixed official PDF

已比较固定会议PDF与arXiv v1的完整图形；目标视觉版本匹配。页面排版平移已在historical_crop_bbox中记录；非图形重绘。实际创作日期未知。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Xiangyu Zhang；481440；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[49, 70, 548, 191]，PDF点、左上角原点。

Figure 3: The diagram of the proposed progressive end-to-end object detection framework. First, the prediction selector
select queries associated with high confidence scores as accepted queries, leaving the rest as noisy queries. Then Relation
information extractor models the relations between noisy queries and their neighbors from accepted predictions. Next, the
queries are fed into the queries updater to be further refined by performing a new local self-attention.

最早核实视觉日期：2022-03-15；arXiv v1：2022-03-15；CVPR 2022 fixed official PDF

已比较固定会议PDF与arXiv v1的完整图形；目标视觉版本匹配。页面排版平移已在historical_crop_bbox中记录；非图形重绘。实际创作日期未知。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Xiangyu Zhang；481440；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[308, 219, 547, 353]，PDF点、左上角原点。

Figure 1: 1a. Sparse RCNN [37] introduces false positives
in crowded scenes. 1b. Our approach can remove those
false positives and ensure each object can be detected only
once. Green boxes indicate true positives while red ones
represent false positives.

最早核实视觉日期：2022-03-30；arXiv v1：2022-03-29；CVPR 2022 fixed official PDF

已阅读会议版图注及正文并逐张查看导出。会议图与对应arXiv历史裁剪像素一致（MAE=0）。v1有符号或细节差异，v2（2022-03-30）首次核实逐像素一致，未套用v1日期。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Wei-Shi Zheng；31913；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[58, 73, 545, 282]，PDF点、左上角原点。

Figure 2. The overview of proposed DMiner, which consists of a Similarity-based Pseudo Label Generating module (SPLG) and a Pixel-
level Group Contrastive Learning module (PGCL). Let M ,M ,M =1 and refer to Eq. (6) to understand the meaning of M.
15 27 31

最早核实视觉日期：2022-03-30；arXiv v1：2022-03-29；CVPR 2022 fixed official PDF

已阅读会议版图注及正文并逐张查看导出。会议图与对应arXiv历史裁剪像素一致（MAE=0）。v1有符号或细节差异，v2（2022-03-30）首次核实逐像素一致，未套用v1日期。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Wei-Shi Zheng；31913；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[309, 240, 545, 345]，PDF点、左上角原点。

Figure 1. Different annotation setup for (a) WSOD, (b) SSOD, (c)
SAOD and (d) SIOD (Ours).

最早核实视觉日期：2022-03-23；arXiv v1：2022-03-23；CVPR 2022 fixed official PDF

已阅读会议版图注及正文并逐张查看导出。会议图与对应arXiv历史裁剪像素一致（MAE=0）。arXiv v1即为最早核实匹配版本。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Yizhou Yu；35607；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[309, 213, 547, 428]，PDF点、左上角原点。

Figure 1. The overall framework of SED. Our model improves the
scale equivalence, which is critical for object detectors, by regu-
larizing the consistency between different-sized images. Further-
more, the inherent False Negative sample noise is alleviated by
self-distillation. A re-weighting strategy is adopted to solve the
severe class imbalance problem. The bird in the left example is a
False Negative sample when the threshold is set to 0.7. The right
example shows the scale inconsistency of different-sized images.

最早核实视觉日期：2022-03-23；arXiv v1：2022-03-23；CVPR 2022 fixed official PDF

已阅读会议版图注及正文并逐张查看导出。会议图与对应arXiv历史裁剪像素一致（MAE=0）。arXiv v1即为最早核实匹配版本。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Yizhou Yu；35607；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[48, 49, 553, 174]，PDF点、左上角原点。

Figure 2. (a) For the COCO dataset, all the images are resized such that the short edge has 800 pixels while the long edge has less than
1333 pixels. For the ImageNet dataset, all the images are resized to 224×224 to calculate the statistics. The scale of object is represented
as the square root of the area. We discuss the typical training input size for ImageNet classification and COCO detection tasks. (b) All the
scores are predicted on COCO minival dataset by the RetinaNet detector with FPN and ResNet 50 backbone, which is trained with 10%
COCO data. The score distance is the absolute difference between the predictions of the image in different sizes. The Y-axis is the average
numberofanchorsperimage. (c)Wepredictpseudo-labelontherestofCOCOtrainingdatawithaconvergedFaster-RCNNdetector(with
FPNandResNet50backbone), trainedwith10%COCOdata. Thelowaveragerecallandprecisionshowthathardpseudo-labelincurmore
noise with False Negative samples.

最早核实视觉日期：2022-03-29；arXiv v1：2022-03-29；CVPR 2022 fixed official PDF

已查看会议固定版与arXiv v1，目标图的构图、元素、连线、文字与颜色一致；存在排版平移或PDF渲染/压缩级别差异，无重绘证据。最早核实公开时间采用v1。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Limin Wang；57767；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[49, 70, 548, 275]，PDF点、左上角原点。

Figure 2. Framework overview. Best viewed in color. We develop the high-level feature alignment into separate task spaces, by applying
the proposed Task-specific Inconsistency Alignment module to both the classification (green part) and localization (blue part) branches
of the baseline detector (gray part). In each branch, the behavioral inconsistency of multiple auxiliary predictors is optimized via the
corresponding inconsistency-aware loss, for essentially bridging the category-wise or boundary-wise margins between domains.

最早核实视觉日期：2022-03-29；arXiv v1：2022-03-29；CVPR 2022 fixed official PDF

已查看会议固定版与arXiv v1，目标图的构图、元素、连线、文字与颜色一致；存在排版平移或PDF渲染/压缩级别差异，无重绘证据。最早核实公开时间采用v1。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Limin Wang；57767；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[309, 211, 546, 337]，PDF点、左上角原点。

Figure 1. Exampled images from PASCAL VOC [10] → Cli-
part [18]. Compared with vanilla detector [28], both UMT [8] and
TIA identify more foreground objects (Row 1), yet deliver lower
as well as higher quality bounding boxes (Row 2), respectively.

最早核实视觉日期：2022-04-06；arXiv v1：2022-04-06；CVPR 2022 fixed official PDF

已比较固定会议PDF与arXiv v1的完整图形；目标视觉版本匹配。页面排版平移已在historical_crop_bbox中记录；非图形重绘。实际创作日期未知。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Yixuan Yuan；16227；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[68, 72, 521, 313]，PDF点、左上角原点。

Figure 3. Overview of our NLTE framework, which includes PIM, MGRM and EAGR. ⃝c is the concatenation operation.

最早核实视觉日期：2022-06；arXiv v1：2022-04-06；CVPR 2022 fixed official PDF

v1与会议版三幅示例内容一致，但子图宽度与间距被调整。已查看全部arXiv历史版本；最早核实包含目标视觉版本的文献为CVF固定会议PDF，官方书目明确出版月June 2022。该月全在S1。月内首次公开日未知，不把会议版新画面追溯为v1日期。历史对照图保留差异，不声称像素相同。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Yixuan Yuan；16227；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[313, 225, 544, 286]，PDF点、左上角原点。

Figure 1. Examples of noisy annotations in Cityscapes dataset.
Miss-annotated samples: The bicycle in (a); the rider and car in
(c). Class-corrupted samples: The rider and bicycle are labeled
as person in (a); the motorcycle is labeled as bicycle in (b).

最早核实视觉日期：2022-03-18；arXiv v1：2022-03-18；CVPR 2022 fixed official PDF

已阅读会议版图注及正文并逐张查看导出。会议图与对应arXiv历史裁剪像素一致（MAE=0）。arXiv v1即为最早核实匹配版本。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Lei Zhang；137335；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[68, 69, 533, 249]，PDF点、左上角原点。

Figure 1. The framework of Class-balanced Pixel-level Self-Labeling (CPSL). The model contains a main segmentation network f and
SEG
its momentum-updated version f′ . The f′ is followed by a self-labeling head f and its momentum version f′ , which projects
SEG SEG SL SL
pixel-wise feature embedding into a class probability vector. The pixel-level self-labeling module produces soft cluster assignment P to
SL
gradually rectify soft pseudo label P . Then the segmentation loss Lt is computed between the prediction map P and the rectified
ST SEG
pseudo label Yˆt. To train the self-labeling head, we randomly sample pixels from each image, and use the memory bank M, which
contains previous batches of pixel features, to augment the current batch. Then we compute the optimal transport assignment Q over
aug
the augmented data by enforcing class balance, and use the assignment of current batch Q to compute the self-labeling loss L .
cur SL

最早核实视觉日期：2022-03-18；arXiv v1：2022-03-18；CVPR 2022 fixed official PDF

已阅读会议版图注及正文并逐张查看导出。会议图与对应arXiv历史裁剪像素一致（MAE=0）。arXiv v1即为最早核实匹配版本。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Lei Zhang；137335；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[48, 70, 290, 226]，PDF点、左上角原点。

Figure 2. The class distribution of the Cityscapes dataset.

最早核实视觉日期：2022-04-15；arXiv v1：2021-12-15；CVPR 2022 fixed official PDF

已核查v1第5页图2：seen/unseen类别框、子图背景和矩阵着色不同。v2第4页匹配会议版，逐像素一致。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Dengxin Dai；23615；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[75, 65, 519, 264]，PDF点、左上角原点。

Figure 2. The pipeline of our proposed ZegFormer for zero-shot semantic segmentation. We first feed N queries and feature maps to a
transformer decoder to generate N segment embeddings. We then feed each segment embedding to a mask projection layer and a semantic
projection layer to obtain a mask embedding and a semantic embedding. Mask embedding is multiplied with the output of pixel decoder
to generate a class-agnostic binary mask, while The semantic embedding is classified by the text embeddings. The text embeddings are
generated by putting the class names into a prompt template and then feeding them to a text encoder of a vision-language model. During
training, only the seen classes are used to train the segment-level classification head. During inference, both the text embeddings of seen
and unseen classes are used for segment-level classification. We can obtain two segment-level classification scores with semantic segment
embeddings and image embeddings. Finally, we fuse the these two classification scores as our final class prediction of segments.

最早核实视觉日期：2022-04-15；arXiv v1：2021-12-15；CVPR 2022 fixed official PDF

已核查v1：seen/unseen及language等文字使用粗斜体；v2改为会议版的正体。v2与会议图逐像素一致，不套用v1日期。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Dengxin Dai；23615；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[310, 208, 541, 447]，PDF点、左上角原点。

Figure 1. ZS3 aims to train a model merely on seen classes and
generalize it to classes that have not been seen in the training (un-
seen classes). Existing methods formulate it as a pixel-level zero-
shot classification problem (b), and use semantic features from a
language model to transfer the knowledge from seen classes to
unseen ones. In contrast, as in (c), we decouple ZS3 into two
sub-tasks: 1) A class-agnostic grouping and 2) A segment-level
zero-shot classification, which enables us to take full advantage of
the pre-trained vision-language model.

最早核实视觉日期：2022-06；arXiv v1：2022-03-05；CVPR 2022 fixed official PDF

v1及v2为Initial pseudo，会议版补为Initial pseudo label。已查看全部arXiv历史版本；最早核实包含目标视觉版本的文献为CVF固定会议PDF，官方书目明确出版月June 2022。该月全在S1。月内首次公开日未知，不把会议版新画面追溯为v1日期。历史对照图保留差异，不声称像素相同。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Bo Du；50901；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[49, 70, 548, 261]，PDF点、左上角原点。

Figure 3. The proposed end-to-end framework for WSSS. We use a Transformer backbone as the encoder to extract feature maps. The
initial pseudo labels are generated with CAM [59] and then refined with the proposed PAR. In the AFA module, we derive the semantic
affinity from MHSA in Transformer blocks. AFA is supervised with the pseudo affinity labels derived from the refined label. Next, we
employ the learned affinity to revise the pseudo labels via random walk propagation [2, 1]. The propagated labels are finally refined with
PAR as the pseudo labels for the segmentation branch.

最早核实视觉日期：2022-03-05；arXiv v1：2022-03-05；CVPR 2022 fixed official PDF

已比较固定会议PDF与arXiv v1的完整图形；目标视觉版本匹配。页面排版平移已在historical_crop_bbox中记录；非图形重绘。实际创作日期未知。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Bo Du；50901；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[313, 271, 540, 456]，PDF点、左上角原点。

Figure 1. (a) Image and the query points (denote with ”⋆”) to vi-
sualize the attention and affinity maps; (b) the self-attention maps
in Transformer blocks only capture coarse semantic-level affin-
ity relations; (c) the learned reliable semantic affinity from self-
attention with our proposed method.

最早核实视觉日期：2022-03-06；arXiv v1：2022-03-06；CVPR 2022 fixed official PDF

已逐页查看会议版与arXiv v1；图区设计一致，保留移动/缩放及PDF压缩差异，不声称像素完全一致。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Wanli Ouyang；81844；快照2026-10-21。

绘图工具：unknown。已读图注与相关正文；未找到该插图实际制作工具声明。

裁剪坐标：[60, 71, 534, 242]，PDF点、左上角原点。

Figure 2. An overview of the proposed multi-class token transformer (MCTformer-V1). It first splits and transforms an input RGB image
into a sequence of patch tokens. We propose to learn C extra class tokens, where C is the number of classes. The C class tokens are
concatenated with patch tokens, with added position embeddings (PE), which then go through consecutive L transformer encoding layers.
Finally, the output C class tokens are used to produce class scores via average pooling. We aggregate the transformer attentions from the
last K layers and multiple heads to generate a final attention map, from which we can extract class-specific object localization maps and a
patch-level pairwise affinity map from the class-to-patch and the patch-to-patch attentions, respectively. The patch-level pairwise affinity
can be used to refine the class-specific transformer attention maps to produce improved object localization maps.

最早核实视觉日期：2022-03-06；arXiv v1：2022-03-06；CVPR 2022 fixed official PDF

已逐页查看会议版与arXiv v1；图区设计一致，保留移动/缩放及PDF压缩差异，不声称像素完全一致。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Wanli Ouyang；81844；快照2026-10-21。

绘图工具：unknown。已读图注与相关正文；未找到该插图实际制作工具声明。

裁剪坐标：[307, 240, 546, 357]，PDF点、左上角原点。

Figure 1. (a) In previous vision transformers [10], only one class
token (red square) is used to aggregate information from patch to-
kens (blue square). The learned patch attentions corresponding
to the class token generate a class-agnostic localization map. (b)
In contrast, the proposed MCTformer uses multiple class tokens
to learn interactions between class tokens and patch tokens. The
learned class-to-patch attentions of different class tokens can pro-
duce class-specific object localization maps.

最早核实视觉日期：2021-12-03；arXiv v1：2021-12-03；CVPR 2022 fixed official PDF

已查看固定PDF及目标图；以最早已核实匹配版本日期作时间代理，不声称实际作图日期。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Nicu Sebe；59898；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[65, 69, 529, 309]，PDF点、左上角原点。

Figure 2. The overview of Entropy-based Uncertainty Modeling and Self-training framework. 1) The model is first trained on the labeled
base set. 2) For the novel images, the saliency maps are generated from a saliency model and high confidence base pixels are ignored. We
then apply clustering to the novel salient part to obtain clustering pseudo-labels. 3) We apply entropy ranking and dynamic reassignment
to split clean and unclean novel images and utilize them to optimize the model with clustering labels and online labels, respectively.

最早核实视觉日期：2021-12-03；arXiv v1：2021-12-03；CVPR 2022 fixed official PDF

v1与会议版的图内布局、颜色、元素一致，正文排版造成图整体下移；分别保存裁剪坐标。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Nicu Sebe；59898；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[317, 226, 535, 474]，PDF点、左上角原点。

Figure 1. Illustration of Novel Class Discovery in Semantic Seg-
mentation(NCDSS).Givenalabeledset(person,horse,motorbike
and car) and a class disjoint unlabeled set (bike and boat), NCDSS
aims at leveraging all the data to segment novel images.

最早核实视觉日期：2022-04-12；arXiv v1：2022-04-12；CVPR 2022 fixed official PDF

已逐页查看会议版与arXiv v1；图区设计一致，保留移动/缩放及PDF压缩差异，不声称像素完全一致。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Jitendra Malik；374128；快照2026-10-21。

绘图工具：unknown。已读图注与相关正文；未找到该插图实际制作工具声明。

裁剪坐标：[60, 71, 534, 184]，PDF点、左上角原点。

Figure 2. Overview of our approach. (a) First, a Pairwise-Affinity Predictor (PA Pred.) is trained to predict pairwise affinities using non-
exhaustive segmentation masks as supervision. (b) Once trained, PA Pred. is used to predict pairwise affinities of the images. A grouping
module is then applied on the predicted pairwise affinity maps to generate pseudo-ground-truth masks. (c) A class-agnostic generic object
proposal network (e.g., class-agnostic Mask R-CNN) is trained end-to-end using a combination of GT and generated pseudo-GT masks.

最早核实视觉日期：2022-04-12；arXiv v1：2022-04-12；CVPR 2022 fixed official PDF

已逐页查看会议版与arXiv v1；图区设计一致，保留移动/缩放及PDF压缩差异，不声称像素完全一致。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Jitendra Malik；374128；快照2026-10-21。

绘图工具：unknown。已读图注与相关正文；未找到该插图实际制作工具声明。

裁剪坐标：[49, 231, 546, 413]，PDF点、左上角原点。

Figure 1. Comparison of GGN with different baselines. (a) an input image from ADE20K [61] (upper row) with the ground truth
object masks overlaid (lower row). Three different approaches: (b) Selective Search (SS) [51], (c) Mask R-CNN [22], and (d) our Generic
Grouping Network (GGN) are applied on the same image to predict the top 100 proposals. The upper images provide all top-100 proposals
predicted by three approaches with false and true positive proposals visualized in red and green boxes, respectively. The lower images
provide only true positive proposals. The number of true positive proposals or ground truth objects are denoted in parentheses. Among 20
ground truth objects, SS recalls only 4, Mask R-CNN detects 5, and GGN retrieves 14. SS is a bottom-up non-parametric approach, thus
has no notion of objectness. Mask R-CNN can make whole object proposals; however it still fails to detect objects that are not seen during
training. Our GGN can predict whole object proposals and generalize to unseen categories.

最早核实视觉日期：2022-03-10；arXiv v1：2022-03-10；CVPR 2022 fixed official PDF

已比较固定会议PDF与arXiv v1的完整图形；目标视觉版本匹配。页面排版平移已在historical_crop_bbox中记录；非图形重绘。实际创作日期未知。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Ming-Ming Cheng；85763；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[314, 207, 541, 357]，PDF点、左上角原点。

Figure 1. Illustration of our proposed training framework for con-
tinual semantic segmentation to avoid catastrophic forgetting. We
design two mechanisms in our method, representation compensa-
tion (RC) module and pooled cube distillation (PCD).

最早核实视觉日期：2022-03-10；arXiv v1：2022-03-10；CVPR 2022 fixed official PDF

已比较固定会议PDF与arXiv v1的完整图形；目标视觉版本匹配。页面排版平移已在historical_crop_bbox中记录；非图形重绘。实际创作日期未知。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Ming-Ming Cheng；85763；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[49, 69, 286, 194]，PDF点、左上角原点。

Figure 5. The impact of the strip pooling (top row) used in
PLOP [27] and the average pooling (bottom row) in our method.

最早核实视觉日期：2022-03-28；arXiv v1：2022-03-28；CVPR 2022 fixed official PDF

v1图2与会议图逐像素一致。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Luc Van Gool；309926；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[49, 66, 548, 183]，PDF点、左上角原点。

Figure 2. Architecture illustration of our non-learnable prototype based nonparametric segmentation model during the training phase.

最早核实视觉日期：2022-03-28；arXiv v1：2022-03-28；CVPR 2022 fixed official PDF

v1图1与会议图逐像素一致。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Luc Van Gool；309926；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[308, 201, 548, 368]，PDF点、左上角原点。

Figure 1. Different sematic segmentation paradigms: (a-b) para-
metric vs (c) nonparametric. Modern segmentation solutions, no
matter using (a) parametric softmax or (b) query vectors for mask
decoding,canbeviewedaslearnableprototypebasedmethodsthat
learn class-wise prototypes in a fully parametric manner. We in-
steadproposeanonparametricscheme(c)thatdirectlyselectssub-
clustercentersofembeddedpixelsasprototypes,andachievesper-
pixel prediction via nonparametric nearest prototype retrieving.

最早核实视觉日期：2022-06；arXiv v1：2022-03-29；CVPR 2022 fixed official PDF

v1的L_LC为单向箭头，v2已改成双向；会议版进一步调整全图比例与标签排布。已查看全部arXiv历史版本；最早核实包含目标视觉版本的文献为CVF固定会议PDF，官方书目明确出版月June 2022。该月全在S1。月内首次公开日未知，不把会议版新画面追溯为v1日期。历史对照图保留差异，不声称像素相同。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Yixuan Yuan；16227；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[49, 57, 548, 198]，PDF点、左上角原点。

Figure 2. The proposed domain adaption framework contains a segmentation net and a SimT. Target images are passed through the
segmentation net f(·) to perform semantic segmentation, and supervision signals are corrected by the learnable SimT. SimT is estimated
w
through the proposed (a) volume regularization, (b) anchor guidance and (c) convex guarantee.

最早核实视觉日期：2022-06；arXiv v1：2022-03-29；CVPR 2022 fixed official PDF

v1及v2与会议版的几何示意比例、标签和图例位置有调整，未强行认作逐像素同一视觉版本。已查看全部arXiv历史版本；最早核实包含目标视觉版本的文献为CVF固定会议PDF，官方书目明确出版月June 2022。该月全在S1。月内首次公开日未知，不把会议版新画面追溯为v1日期。历史对照图保留差异，不声称像素相同。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Yixuan Yuan；16227；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[48, 64, 289, 161]，PDF点、左上角原点。

Figure 1. Geometric illustration of SimT with the setting of 3
pseudo label classes and 5 ground truth label classes. Note that
coordinates (x,y,z) of any points in blue triangle satisfy condi-
tions of x + y + z = 1 and 0 < x,y,z < 1.

最早核实视觉日期：2022-06；arXiv v1：2022-03-21；CVPR 2022 fixed official PDF

v1及v2的分割模型S(.)使用衬线字体，会议版该标签改为无衬线字体。已查看全部arXiv历史版本；最早核实包含目标视觉版本的文献为CVF固定会议PDF，官方书目明确出版月June 2022。该月全在S1。月内首次公开日未知，不把会议版新画面追溯为v1日期。历史对照图保留差异，不声称像素相同。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Xiangyu Zhang；481440；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[72, 70, 523, 207]，PDF点、左上角原点。

Figure 3. Flowchart of the proposed single-stage SASS method, which is realized by incorporating an auxiliary branch into the traditional
segmentation model S(·). During training, the predicted masks P are split into the labeled and unlabeled parts, which are supervised by
the segmentation loss L and the tree energy loss L , respectively. To obtain pseudo labels for unlabeled pixels, the Tree Affinity
seg tree
Generation procedure (Eqs. 3-5) first utilizes the color information I and semantic features F to generate the low-level and high-level
affinity matrices Alow,Ahigh. Then the Cascaded Filtering operation (Eqs. 6-7) converts the network predictions P into soft pseudo labels
Y˜. During testing, the auxiliary branch is removed to avoid extra computational costs.

最早核实视觉日期：2022-06；arXiv v1：2022-03-21；CVPR 2022 fixed official PDF

v1及v2的子图标题为衬线字体，会议版改为无衬线字体，且v2块形状也不同。已查看全部arXiv历史版本；最早核实包含目标视觉版本的文献为CVF固定会议PDF，官方书目明确出版月June 2022。该月全在S1。月内首次公开日未知，不把会议版新画面追溯为v1日期。历史对照图保留差异，不声称像素相同。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Xiangyu Zhang；481440；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[320, 214, 535, 516]，PDF点、左上角原点。

Figure1. IllustrationofcurrentSASSapproaches. S andE denote
the segmentation and auxiliary models, respectively. Our method
leverages the minimum spanning trees (MSTs) to capture both
low-level and high-level affinities to generate soft pseudo labels,
performing online self-training.

最早核实视觉日期：2022-03-17；arXiv v1：2022-03-17；CVPR 2022 fixed official PDF

已查看固定PDF及目标图；以最早已核实匹配版本日期作时间代理，不声称实际作图日期。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Lei Zhang；137335；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[72, 69, 525, 219]，PDF点、左上角原点。

Figure2. ArchitectureofourproposedTATTnetworkforSTISR.TPGB,TPGandSRBareshortfortextpriorguidedblocks,TPGenerator
and Sequential-Recurrent Blocks, respectively, while ⊕ means the element-wise addition.

最早核实视觉日期：2022-03-17；arXiv v1：2022-03-17；CVPR 2022 fixed official PDF

已查看固定PDF及目标图；以最早已核实匹配版本日期作时间代理，不声称实际作图日期。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Lei Zhang；137335；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[309, 230, 546, 390]，PDF点、左上角原点。

Figure 1. SR recovery of different models on rotated and curve-
shaped text images. ‘R’, ‘P’ and ‘S’ stand for recognition, PSNR
and SSIM results. Characters in red are missing or wrong.

最早核实视觉日期：2021-11-11；arXiv v1：2021-07-02；CVPR 2022 fixed official PDF

v1/v2流程图为旧版；v3图1与会议版逐像素一致；v4未改变。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Lei Zhang；137335；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[91, 71, 502, 243]，PDF点、左上角原点。

Figure 1. An overview of the proposed SISR method and the corresponding EM algorithm. A probabilistic model is constructed to depict the generation
process of the observed LR image, which mainly involves two groups of parameters, including the latent variable z and the model parameters {α,L,λ}.
AMonteCarloEMalgorithmisdesignedtoalternatelyupdatethemintheE-StepandM-Step,respectively.

最早核实视觉日期：2021-11-11；arXiv v1：2021-07-02；CVPR 2022 fixed official PDF

v1/v2旧图2含噪声方差图，构图不同；v3已换为会议图3三面板，相比会议排版纵坐标上移2.834pt；对应坐标逐图比对。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Lei Zhang；137335；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[308, 271, 546, 358]，PDF点、左上角原点。

Figure3. IllustrationofcamerasensornoiseofCase2. Fromlefttoright:
(a) clean image; (b) simulated noisy image with camera sensor noise; (c)
absoluteresidual(ornoise)between(a)and(b).

最早核实视觉日期：2021-12-05；arXiv v1：2021-12-05；CVPR 2022 fixed official PDF

固定会议与v1两张图像素一致（MAE=0）；已读图注正文与Google Research作者单位，完整300dpi导出已目检。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Peyman Milanfar；37874；快照2026-10-22。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[308, 75, 547, 180]，PDF点、左上角原点。

Figure 2. Diagram describing our dual-network architecture. The
initial predictor produces the deterministic candidate for the de-
noiser network, which then models the residual.

最早核实视觉日期：2021-12-05；arXiv v1：2021-12-05；CVPR 2022 fixed official PDF

固定会议与v1两张图像素一致（MAE=0）；已读图注正文与Google Research作者单位，完整300dpi导出已目检。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Peyman Milanfar；37874；快照2026-10-22。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[51, 73, 286, 204]，PDF点、左上角原点。

Figure 3. Output of the initial predictor and multiple samples
generated from it. We see that the over-smoothed initial prediction
lacking texture is “corrected” by the stochastic sampler, producing
crisp and diverse final reconstructions. The residual (top right)
shows the the difference between reference and initial prediction.

最早核实视觉日期：2022-02-15；arXiv v1：2022-02-15；CVPR 2022 fixed official PDF

固定会议PDF与arXiv v1的目标图已对照；构图、配色、标签和连线一致。已在固定PDF首页及公开作者引用主页核对作者身份和机构。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Haoqiang Fan；13604；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[49, 70, 548, 203]，PDF点、左上角原点。

Figure 4. The overview architecture of the proposed method. Given an LR image y, we first estimate the degradation kernel kˆ , and involve
it in the deep constrained least squares (DCLS) convolution in the feature domain. The deblurred features R(cid:98) are then concatenated with
primitive features G y to restore the clean HR image x through a dual-path attention network (DPAN).

最早核实视觉日期：2022-02-15；arXiv v1：2022-02-15；CVPR 2022 fixed official PDF

固定会议PDF与arXiv v1的目标图已对照；构图、配色、标签和连线一致。已在固定PDF首页及公开作者引用主页核对作者身份和机构。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Haoqiang Fan；13604；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[310, 72, 545, 163]，PDF点、左上角原点。

Figure 2. Kernel reformulation examples. The top row and middle
row are the LR images and the corresponding primitive kernels.
The bottom row is the reformulated kernels.

最早核实视觉日期：2022-03-17；arXiv v1：2022-03-17；CVPR 2022 fixed official PDF

固定会议PDF与arXiv v1目标图按对应坐标逐图比对；像素均差为0；arXiv v1与已知会议投稿均在指定阶段。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Lei Zhang；137335；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[329, 341, 509, 435]，PDF点、左上角原点。

Figure 6. Overall learning pipeline of the proposed LDL method.

最早核实视觉日期：2022-03-17；arXiv v1：2022-03-17；CVPR 2022 fixed official PDF

固定会议PDF与arXiv v1目标图按对应坐标逐图比对；像素均差为0；arXiv v1与已知会议投稿均在指定阶段。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Lei Zhang；137335；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[309, 231, 547, 435]，PDF点、左上角原点。

Figure 1. Three representative types of SISR regions generated by
ESRGAN [40]. For each example, the left is an LR patch and the
right is its GAN-SR result. Type A patches represent regions that
are easy to super-resolve, e.g., smooth and large-scale structural
areas, where the main structures are preserved in the LR input. In
contrast, patches of type B and type C are with fine-scale details,
which are hard to be faithfully restored due to the signal alias-
ing in the LR inputs. The results of texture-like type B patches
are perceptually realistic despite the pixel-wise differences to the
ground-truth, since the patterns are naturally irregular with weak
priorsforobservers. However,theresultsoftypeCpatchesexhibit
perceptually unpleasant visual artifacts since the overshoot pixels
and distorted structures are sensitive to human perception.

最早核实视觉日期：2022-05-07；arXiv v1：2022-05-07；CVPR 2022 fixed official PDF

固定会议PDF与arXiv v1的目标图已对照；构图、配色、标签和连线一致。已在固定PDF首页及公开作者引用主页核对作者身份和机构。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Liang Lin；51769；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[50, 70, 548, 299]，PDF点、左上角原点。

Figure 5. The proposed framework of our DADA. It includes a source branch and a target branch, both of which have an LR-HR-LR
reconstruction structure. In their up-sampling networks, a Domain-Invariant Attention (DIA) module provides component attentive masks
by a pre-trained CDC model. For InterAA, two LR images from different domains are fed into the up-sampling network for adversarial
adaptation in one branch. For IntraAA, one LR image is respectively fed into up-sampling networks in source and target branches for the
adversarial adaptation. For testing, the trained up-sampling network in target branch is utilized for inference.

最早核实视觉日期：2022-05-07；arXiv v1：2022-05-07；CVPR 2022 fixed official PDF

固定会议PDF与arXiv v1的目标图已对照；构图、配色、标签和连线一致。已在固定PDF首页及公开作者引用主页核对作者身份和机构。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Liang Lin；51769；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[309, 219, 555, 415]，PDF点、左上角原点。

Figure 1. Comparison of real SR results across different trained
models from different camera data in the DRealSR dataset [18].
The Difference Map denotes the absolute difference between
ground-truth HR and SR image (The brightness in the map reflects
the magnitude of the difference).

最早核实视觉日期：2021-06；arXiv v1：2020-12-01；CVPR 2021 fixed official PDF

已逐图对照arXiv v1与CVPR固定PDF。DASR图1会议版移除左侧统计图；DDDNet图5标签位置及构图比例有改动。其余图同样存在缩放/排版变化，保守使用CVF书目明确的2021-06固定版公开月份，不将重排或重绘版本倒溯至v1。真实作图日期未知；v1至会议公开均处S1。作者单位/邮箱已核对；Shuhang Gu的ETH历史任职由本人主页佐证。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Hongdong Li；33685；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[75, 55, 521, 172]，PDF点、左上角原点。

Figure 5. The architecture of the proposed DDDNet. With the input DP image pair, the DepthNet first estimates a coarse inverse depth map
Dˆ . Then, we combine the coarse and blurred inverse depth map with B , and feed them to the DeblurNet. The deblurred image and
c {L,R}
estimated inverse depth map are fed to our DP simulator to synthesize a DP image pair. The synthesized DP images are compared to the
inputs using our reblur loss (Eq. (11)), to regularise the DDDNet in training. Note that, the ground truth inverse depth maps and sharp
images are used as supervision signals in training. (Best viewed in colour on screen.)

最早核实视觉日期：2021-06；arXiv v1：2020-12-01；CVPR 2021 fixed official PDF

已逐图对照arXiv v1与CVPR固定PDF。DASR图1会议版移除左侧统计图；DDDNet图5标签位置及构图比例有改动。其余图同样存在缩放/排版变化，保守使用CVF书目明确的2021-06固定版公开月份，不将重排或重绘版本倒溯至v1。真实作图日期未知；v1至会议公开均处S1。作者单位/邮箱已核对；Shuhang Gu的ETH历史任职由本人主页佐证。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Hongdong Li；33685；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[64, 73, 269, 154]，PDF点、左上角原点。

Figure 2. Formation of the two image I and I from two points
L R
C and C on the lens plane may be thought of as pin-hole pro-
L R
jections (with centres C L and C R and focal-length F ) of the vir-
tual image formed by the lens.

最早核实视觉日期：2022-03-22；arXiv v1：2022-03-22；CVPR 2022 fixed official PDF

固定会议PDF与arXiv v1的目标图已对照；构图、配色、标签和连线一致。已在固定PDF首页及公开作者引用主页核对作者身份和机构。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Yanning Zhang；20703；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[308, 72, 557, 194]，PDF点、左上角原点。

Figure 5. The proposed model architecture for IRP prediction.

最早核实视觉日期：2022-03-22；arXiv v1：2022-03-22；CVPR 2022 fixed official PDF

固定会议PDF与arXiv v1的目标图已对照；构图、配色、标签和连线一致。已在固定PDF首页及公开作者引用主页核对作者身份和机构。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Yanning Zhang；20703；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[308, 230, 557, 398]，PDF点、左上角原点。

Figure 1. We show how the proposed concept IRP differs from
the traditional image metric visual quality. Given a sequence of
images captured under varying exposures in dynamic scenes (first
row), both IQA metric NIQE [23] and human observers favorite
either (a) or (b) as relatively good shots, but they do not neces-
sarily lead to better restored results (second row). As a compari-
son, a perceptually poor image (c) leads to the best restored image
quality. The result indicates the necessity of developing the IRP
metric which predicts if the restored image quality will be good
even before it is processed. All images are processed by the same
restoration model MPR [40] to ensure fairness.

最早核实视觉日期：2021-04-21；arXiv v1：2021-04-21；CVPR 2021 fixed official PDF

已核对固定会议PDF与arXiv v1完整图页，未见目标图重绘；微小像素差源于PDF渲染。作者姓名与固定PDF单位、公开Scholar单位相符。已逐张查看300dpi导出、确认完整。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Kyoung Mu Lee；57875；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[49, 72, 546, 226]，PDF点、左上角原点。

Figure 4: Overall organization of the proposed SRWarp model. More detailed architectures are described in our supplementary material. Black regions outside the warped image I_SR represent void pixels that are ignored.

最早核实视觉日期：2021-04-21；arXiv v1：2021-04-21；CVPR 2021 fixed official PDF

已核对固定会议PDF与arXiv v1完整图页，未见目标图重绘；微小像素差源于PDF渲染。作者姓名与固定PDF单位、公开Scholar单位相符。已逐张查看300dpi导出、确认完整。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Kyoung Mu Lee；57875；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[308, 70, 546, 184]，PDF点、左上角原点。

Figure 2: Concept of the generalized SR. Dotted lines represent the upscale image pyramid in xyz-coordinate. While ×s super-resolved images from conventional methods (red) exist on the plane z = s only, our results (blue) exists on any arbitrary cutting surfaces of the pyramid. Note that the aspect ratio on the plane can be varied as well.

最早核实视觉日期：2021-06；arXiv v1：2020-04-02；CVPR 2021 fixed official PDF

已逐图对照arXiv v1与CVPR固定PDF。DASR图1会议版移除左侧统计图；DDDNet图5标签位置及构图比例有改动。其余图同样存在缩放/排版变化，保守使用CVF书目明确的2021-06固定版公开月份，不将重排或重绘版本倒溯至v1。真实作图日期未知；v1至会议公开均处S1。作者单位/邮箱已核对；Shuhang Gu的ETH历史任职由本人主页佐证。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Shuhang Gu；21404；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[133, 73, 459, 213]，PDF点、左上角原点。

Figure 2: Illustration of our DASR framework. DASR firstly trains a down-sampling network (DSN) which aims to generate
LR images y g in the real LR domain y r from HR images xr : y g = DS N (xr ). Then, DASR take the domain-gap between
generated LR images y g and real-world LR images y r into consideration, utilize the generated LR-HR pairs {y g , xr }, domain
distance maps w and real-world LR images y r to train super-resolution network (SRN).

最早核实视觉日期：2021-06；arXiv v1：2020-04-02；CVPR 2021 fixed official PDF

已逐图对照arXiv v1与CVPR固定PDF。DASR图1会议版移除左侧统计图；DDDNet图5标签位置及构图比例有改动。其余图同样存在缩放/排版变化，保守使用CVF书目明确的2021-06固定版公开月份，不将重排或重绘版本倒溯至v1。真实作图日期未知；v1至会议公开均处S1。作者单位/邮箱已核对；Shuhang Gu的ETH历史任职由本人主页佐证。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Shuhang Gu；21404；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[312, 258, 542, 423]，PDF点、左上角原点。

Figure 1: Visualization of the domain-gap between the gen-
erated LR images by different methods and LR images in
the AIM [40] dataset. More details can be found in our
main text.

最早核实视觉日期：2021-04-01；arXiv v1：2021-04-01；CVPR 2021 fixed official PDF

已查看会议固定版与arXiv v1，目标图的构图、元素、连线、文字与颜色一致；存在排版平移或PDF渲染/压缩级别差异，无重绘证据。最早核实公开时间采用v1。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Thomas Brox；249461；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[50, 70, 287, 170]，PDF点、左上角原点。

Figure 3. We use a hierarchy of networks operating on input
patches of different resolutions (including the global one) to pro-
duce multiple shape reconstruction variants. Those are fused by
simple averaging to yield the final reconstruction.

最早核实视觉日期：2021-04-01；arXiv v1：2021-04-01；CVPR 2021 fixed official PDF

已查看会议固定版与arXiv v1，目标图的构图、元素、连线、文字与颜色一致；存在排版平移或PDF渲染/压缩级别差异，无重绘证据。最早核实公开时间采用v1。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Thomas Brox；249461；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[311, 244, 545, 455]，PDF点、左上角原点。

Figure 1. We employ a hierarchical shape prior to enable recom-
bination of partial shapes observed during training. This signif-
icantly improves generalization compared to conventional global
shape priors.

最早核实视觉日期：2020-12-21；arXiv v1：2020-12-21；CVPR 2021 fixed official PDF

固定PDF图注、正文引用段落及作者单位已核验。导出完整图区域保留文字符号，300dpi与历史裁剪已逐张目检。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Bastian Leibe；44197；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[49, 70, 548, 182]，PDF点、左上角原点。

Figure 2: Overview of the proposed approach. Given a single RGB image, our model detects object centers as key-points in a heatmap Y .
The network directly predicts shape exemplars z and 9-DoF bounding boxes jointly for all objects in the scene. The collision loss L
coll
favors non-intersecting reconstructions. Our method predicts lightweight, realistic and physically plausible reconstructions in a single pass.

最早核实视觉日期：2021-06；arXiv v1：2020-12-21；CVPR 2021 fixed official PDF

已核查arXiv v1(2020-12-21)、v2(2021-04-07)、v3(2021-06-21)：三者M的文字字形/渲染均不同于固定会议图；概念构图相同。故保守使用CVF官方Bibtex公开月份2021-06，明确月份精度，未推断确切首发日。三版裁剪及会议导出已查看。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Bastian Leibe；44197；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[58, 73, 280, 142]，PDF点、左上角原点。

Figure 4: Shape selection. We compare one-hot encoding (hard
labels, left) for supervising the shape selection problem with soft
labels (right) which allow for multiple shape predictions at the
same time and are based on geometric similarity, specifically the
Euclidian distance between SDF shape representations.

最早核实视觉日期：2018-11-26；arXiv v1：2018-11-26；CVPR 2019 fixed official PDF

已查看会议固定版与arXiv v1，目标图的构图、元素、连线、文字与颜色一致；存在排版平移或PDF渲染/压缩级别差异，无重绘证据。最早核实公开时间采用v1。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Tatsuya Harada；24085；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[49, 72, 548, 197]，PDF点、左上角原点。

Figure 2. Architecture of the proposed method. The main point of our method is the use of discrimination loss to learn priors of views.
While the discriminator aims to minimize discrimination loss, the encoder and decoders try to maximize it using a gradient reversal layer.

最早核实视觉日期：2018-11-26；arXiv v1：2018-11-26；CVPR 2019 fixed official PDF

已查看会议固定版与arXiv v1，目标图的构图、元素、连线、文字与颜色一致；存在排版平移或PDF渲染/压缩级别差异，无重绘证据。最早核实公开时间采用v1。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Tatsuya Harada；24085；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[309, 214, 546, 384]，PDF点、左上角原点。

Figure 1. When a 3D reconstructor is trained using only a single
viewperobject,becauseofambiguityinthe3Dshapeofanobject,
itreconstructsashape whichonly fitsthe observedview andlooks
incorrect from unobserved viewpoints (upper). By introducing a
discriminator that learns prior knowledge of correct views, the re-
constructor is able to generate a shape that is viewed as reasonable
from any viewpoint (lower).

最早核实视觉日期：2021-04-01；arXiv v1：2021-04-01；CVPR 2021 fixed official PDF

已查看会议固定版与arXiv v1，目标图的构图、元素、连线、文字与颜色一致；存在排版平移或PDF渲染/压缩级别差异，无重绘证据。最早核实公开时间采用v1。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Xiaowei Zhou；25061；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[49, 41, 548, 280]，PDF点、左上角原点。

Figure 2. NeuralRecon architecture. NeuralRecon predicts TSDF with a three-level coarse-to-fine approach that gradually increases the
density of sparse voxels. Key-frame images in the local fragment are first passed through the image backbone to extract the multi-level
features. These image features are later back-projected along each ray and aggregated into a 3D feature volume Fl, where l represents the
t
level index. At the first level (l = 1), a dense TSDF volume S1 is predicted. At the second and third levels, the upsampled Sl−1 from the
t t
last level is concatenated with Fl and used as the input for the GRU Fusion and MLP modules. A feature volume defined in the world
t
frame is maintained at each level as the global hidden state of the GRU. At the last level, the output Sl is used to replace corresponding
t
voxels in the global TSDF volume Sg, yielding the final reconstruction at time t.
t

最早核实视觉日期：2021-04-01；arXiv v1：2021-04-01；CVPR 2021 fixed official PDF

已查看会议固定版与arXiv v1，目标图的构图、元素、连线、文字与颜色一致；存在排版平移或PDF渲染/压缩级别差异，无重绘证据。最早核实公开时间采用v1。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Xiaowei Zhou；25061；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[308, 204, 548, 351]，PDF点、左上角原点。

Figure1. Comparisonbetweendepth-based3Dreconstruction
methods and the proposed method. In depth-based methods,
key-frame depths are estimated separately from each key frame,
and later fused into a TSDF volume. In the proposed method, the
TSDF volume is directly predicted with all the key frames in a
local window. This design leads to a much more coherent recon-
struction and real-time speed.

最早核实视觉日期：2018-12-10；arXiv v1：2018-12-10；CVPR 2019 fixed official PDF

已查看会议固定版与arXiv v1，目标图的构图、元素、连线、文字与颜色一致；存在排版平移或PDF渲染/压缩级别差异，无重绘证据。最早核实公开时间采用v1。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Andreas Geiger；110140；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[52, 75, 287, 250]，PDF点、左上角原点。

Figure2: MultiresolutionIsoSurfaceExtraction: Wefirst
mark all points at a given resolution which have already
been evaluated as either occupied (red circles) or unoccu-
pied (cyan diamonds). We then determine all voxels that
have both occupied and unoccupied corners and mark them
as active (light red) and subdivide them into 8 subvoxels
each. Next, we evaluate all new grid points (empty circles)
that have been introduced by the subdivision. The previous
two steps are repeated until the desired output resolution is
reached. Finally we extract the mesh using the marching
cubes algorithm [44], simplify and refine the output mesh
using first and second order gradient information.

最早核实视觉日期：2019-04-30；arXiv v1：2018-12-10；CVPR 2019 fixed official PDF

v1(2018-12-10)图1(c)的Mesh长椅实例与固定会议版不同；v2(2019-04-30)替换为与会议版相同长椅，按v2日期。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Andreas Geiger；110140；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[311, 241, 544, 364]，PDF点、左上角原点。

Figure 1: Overview: Existing 3D representations discretize
the output space differently: (a) spatially in voxel represen-
tations, (b) in terms of predicted points, and (c) in terms of
vertices for mesh representations. In contrast, (d) we pro-
pose to consider the continuous decision boundary of a clas-
sifier f (e.g., a deep neural network) as a 3D surface which
θ
allows to extract 3D meshes at any resolution.

最早核实视觉日期：2022-04-19；arXiv v1：2022-04-19；CVPR 2022 fixed official PDF

已查看固定会议PDF、图注和相关正文，核对唯一/最早v1目标图；源裁图与v1逐像素一致，MAE=0。保存v1及对应完整裁图。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Cristian Sminchisescu；24653；快照2026-10-22。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[58, 70, 535, 182]，PDF点、左上角原点。

Figure 2. Overview of our method. The feature extractor network G produces pixel-aligned features z from an input image I for all points
x
in space x. The implicit signed distance function network f computes the distance d to the closest surface given a point and its feature.
Additionally f returns albedo colors a defined for surface points. The shading network s predicts the shading for surface points given its
surface normal n , as well as illumination l. On the right we show the reconstruction of geometry and albedo colors, and the shaded 3D
x
geometry.

最早核实视觉日期：2022-04-19；arXiv v1：2022-04-19；CVPR 2022 fixed official PDF

已查看固定会议PDF、图注和相关正文，核对唯一/最早v1目标图；源裁图与v1逐像素一致，MAE=0。保存v1及对应完整裁图。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Cristian Sminchisescu；24653；快照2026-10-22。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[308, 332, 544, 395]，PDF点、左上角原点。

Figure 8. Failure cases. Wide clothing is under-represented in
our dataset and this can be addressed with more diverse training.
Complex poses can lead to missing body parts. The back-side
sometimes mismatches the front (subject is wearing a hood).

最早核实视觉日期：2022-04-07；arXiv v1：2022-04-07；CVPR 2022 fixed official PDF

固定PDF图注、正文引用段落及作者单位已核验。导出完整图区域保留文字符号，300dpi与历史裁剪已逐张目检。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Shubham Tulsiani；13507；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[78, 72, 510, 168]，PDF点、左上角原点。

Figure 2. Approach Overview. We first pre-train a reconstruction model using multi-view renderings of synthetic data. We then self-train
category-specific models on diverse image collections in-the-wild with only foreground mask annotations. We finally distill the learned
models from prior training stages into a unified reconstruction model.

最早核实视觉日期：2022-04-07；arXiv v1：2022-04-07；CVPR 2022 fixed official PDF

固定PDF图注、正文引用段落及作者单位已核验。导出完整图区域保留文字符号，300dpi与历史裁剪已逐张目检。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Shubham Tulsiani；13507；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[310, 72, 543, 155]，PDF点、左上角原点。

Figure 8. Failure modes. Our model is not robust to occlusion
and does not perform well on some difficult categories.

最早核实视觉日期：2019-02-26；arXiv v1：2019-02-26；CVPR 2019 fixed official PDF

已比较arXiv v1（2019-02-26）与固定会议图2，模块、标签、连线及配色一致；轻微像素差异来自栅格压缩/渲染。作者Shenghua Gao在文中单位ShanghaiTech，与公开主页履历一致。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Shenghua Gao；29917；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[75, 72, 520, 278]，PDF点、左上角原点。

Figure 2: Network architecture. In the first stage, the network takes a single RGB image as input, and predicts a planar/non-
planar segmentation mask and pixel-level embeddings. Then, an efficient mean shift clustering algorithm is applied to
generate plane instances. In the second stage, we estimate parameter of each plane by considering both pixel-level and
instance-level geometric consistencies.

最早核实视觉日期：2019-04-24；arXiv v1：2019-02-26；CVPR 2019 fixed official PDF

已查看v1(2019-02-26)、v2(03-04)、v3(04-24)。三行失败案例的行间留白和文字排布在早期版本不同，v3才匹配会议排布；图像内容相同，轻微像素差异来自栅格压缩/渲染。使用v3日期。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Shenghua Gao；29917；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[305, 186, 548, 377]，PDF点、左上角原点。

Figure 7: Failure cases.

最早核实视觉日期：2022-04-23；arXiv v1：2022-04-23；CVPR 2022 fixed official PDF

已查看固定PDF及目标图；以最早已核实匹配版本日期作时间代理，不声称实际作图日期。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Matthias Zwicker；17011；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[49, 70, 287, 137]，PDF点、左上角原点。

Figure 2. Overview of our method.

最早核实视觉日期：2022-04-23；arXiv v1：2022-04-23；CVPR 2022 fixed official PDF

已查看固定PDF及目标图；以最早已核实匹配版本日期作时间代理，不声称实际作图日期。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Matthias Zwicker；17011；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[49, 71, 287, 190]，PDF点、左上角原点。

Figure 4. Current methods can merely search similar regions at a
specific location q(cid:48) in the learned prior space by optimizing con-
l
dition f(cid:48). While we can search over the whole prior space by
l
simultaneously optimizing q(cid:48) and f(cid:48).
l l

最早核实视觉日期：2022-05-12；arXiv v1：2022-05-12；arXiv v1 fixed source

固定arXiv v1原件；会议视觉与早期版本存在差异，未给新版图套用v1日期。TARS已查看v1/v2及会议版；PassionSR已查看v1/v2/v3及会议版；具体差异保留在review页图中。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Deepak Pathak；37267；快照2026-10-21。

绘图工具：unknown。已读图注与相关正文；未找到该插图实际制作工具声明。

裁剪坐标：[68, 70, 526, 194]，PDF点、左上角原点。

Figure 2. Overview of TARS: Given a single image, we first map a 3D point in object space to a higher-dimensional canonical space using
our learned topologically-aware deformation field. The canonical point is then mapped to its SDF value using the Canonical Shape Generator
module. We leverage an LSTM-based differentiable renderer to guide the learning of deformation and signed distance fields.

最早核实视觉日期：2022-05-12；arXiv v1：2022-05-12；arXiv v1 fixed source

固定arXiv v1原件；会议视觉与早期版本存在差异，未给新版图套用v1日期。TARS已查看v1/v2及会议版；PassionSR已查看v1/v2/v3及会议版；具体差异保留在review页图中。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Deepak Pathak；37267；快照2026-10-21。

绘图工具：unknown。已读图注与相关正文；未找到该插图实际制作工具声明。

裁剪坐标：[61, 139, 533, 323]，PDF点、左上角原点。

Figure 1. Given an unpaired image collection (with known camera poses) of an object category at training time, our approach learns to: (a)
reconstruct the underlying 3D given only a single image at test time, and (b) model dense 3D correspondences across category instances.
The learned correspondence field is articulation-aware, topologically-aware and inherently captures structural properties of the category,
enabling the task of unsupervised texture transfer. Videos and code at https://shivamduggal4.github.io/tars-3D/

最早核实视觉日期：2024-03-28；arXiv v1：2024-03-28；CVPR 2024 fixed official PDF

已核对固定会议PDF与arXiv v1完整图页，未见目标图重绘；微小像素差源于PDF渲染。作者姓名与固定PDF单位、公开Scholar单位相符。已逐张查看300dpi导出、确认完整。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Robby T. Tan；18517；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[67, 72, 535, 302]，PDF点、左上角原点。

Figure 2. (a) Class-Aware Teacher (CAT) consists of: a student-teacher network; Inter-Class Relation module (ICRm), which estimates
inter-class biases; Class-Relation Augmentation, which augments images to reduce the inter-class biases by mixing the cropped instances of
related classes; and Inter-Class Loss, which emphasises the loss on highly misclassified minority classes. (b) Class-Relation Augmentation
demonstrated on majority (Car) and minority (Bus) classes.

最早核实视觉日期：2024-03-28；arXiv v1：2024-03-28；CVPR 2024 fixed official PDF

已核对固定会议PDF与arXiv v1完整图页，未见目标图重绘；微小像素差源于PDF渲染。作者姓名与固定PDF单位、公开Scholar单位相符。已逐张查看300dpi导出、确认完整。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Robby T. Tan；18517；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[308, 254, 558, 382]，PDF点、左上角原点。

Figure 1. Performance of Class-Aware Teacher (CAT). AT [29]
(top left), with Inter-Class Loss, ICL, (top-right), with Class Rela-
tion Augmentation, CRA, (bottom-left), and CAT (bottom-right).
CAT is able to address misclassification and false positives, blue
and red boxes, respectively, in minority classes such as ’train’. The
combination of ICL and CRA further boosts performance by re-
ducing the number of false positives shown as pink boxes.

最早核实视觉日期：2025-06-03；arXiv v1：2025-06-03；CVPR 2025 fixed official PDF

已查看固定PDF及目标图；以最早已核实匹配版本日期作时间代理，不声称实际作图日期。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Zheng-Jun Zha；30479；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[58, 70, 555, 305]，PDF点、左上角原点。

Figure 4. Overview of our method. Motivated by the observation that certain feature channels learned from the source domain negatively
impact cross-domain performance, we propose suppressing and pruning sensitive channels while adapting invariant ones. We introduce
sensitivity-guided channel pruning, which quantifies the importance of each feature channel based on its image-level and instance-level
sensitivity to domain discrepancies. We then apply a weighted sparsity loss to the learnable parameters, promoting the pruning of channels
with higher sensitivity based on the predefined threshold and ratio. To prevent the loss of potentially useful channels from early pruning,
we introduce a stochastic channel reactivation mechanism.

最早核实视觉日期：2025-06-03；arXiv v1：2025-06-03；CVPR 2025 fixed official PDF

已查看固定PDF及目标图；以最早已核实匹配版本日期作时间代理，不声称实际作图日期。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Zheng-Jun Zha；30479；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[316, 258, 555, 368]，PDF点、左上角原点。

Figure 1. We consider continual test-time adaptive object detec-
tion, which adapts a source detector to changing environments
during inference. By focusing adaptation efforts on invariant
parts while pruning sensitive parts, we improve computational ef-
ficiency, which is crucial for resource-constrained scenarios like
autonomous driving or UAV.

最早核实视觉日期：2024-10-15；arXiv v1：2024-10-15；CVPR 2025 fixed official PDF

已核查v1第4页图2、v2第3页图2及会议PDF。构图、标签、颜色、连线相同，仅版面缩放；最早匹配v1（2024-10-15）。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Jiankang Deng；26964；快照2026-10-22。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[106, 70, 501, 177]，PDF点、左上角原点。

Figure 2. During imbalanced object detection, the model makes more frequent class detections like hat and less rare class detections like
tiara both of which have strong upper location bias. FRACAL utilises fractal dimension and debiases the logits both in the frequency and
space axes, making fewer hat detections and more tiara detections that are both evenly spread in image space.

最早核实视觉日期：2025-03-05；arXiv v1：2024-10-15；CVPR 2025 fixed official PDF

已核查v1、v2及固定会议PDF。v1图1为横向并排；v2（2025-03-05）改为上下布局且与会议图逐像素一致，采用v2日期。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Jiankang Deng；26964；快照2026-10-22。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[318, 214, 555, 384]，PDF点、左上角原点。

Figure 1. Previous works used class information p (y ) to align
s
the learned source distribution p (y , u|x) with the balanced target
s
distribution p (y , u|x), without considering the space u and class
t
y relationship i.e. p (y , u). FRACAL captures p (y , u), using
s s
the fractal dimension, and embeds fractal margins during infer-
ence, aligning the learned distribution p (y , u|x) with the target
s
p (y , u|x) better than previous works.
t

最早核实视觉日期：2024-03-15；arXiv v1：2024-03-15；CVPR 2024 fixed official PDF

已核对固定会议PDF与arXiv v1完整图页，未见目标图重绘；微小像素差源于PDF渲染。作者姓名与固定PDF单位、公开Scholar单位相符。已逐张查看300dpi导出、确认完整。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Jianfei Cai；40906；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[54, 72, 547, 226]，PDF点、左上角原点。

Figure 2. Overview of our proposed open-ended object detection model, GenerateU, which comprises two major components: an object
detector and a language model. We compare two training strategies: (Top) We incorporate the class-agnostic DETR (with frozen image
encoder) into a pre-trained and frozen Multimodal Large Language Model (including Adaptor and Language Model), to facilitate a smooth
transfer of knowledge from the language domain to object detection; (Bottom) We activate the image encoder and language model as
trainable components, taking an end-to-end approach to seamlessly integrate region-level understanding into the language model.

最早核实视觉日期：2024-03-15；arXiv v1：2024-03-15；CVPR 2024 fixed official PDF

已核对固定会议PDF与arXiv v1完整图页，未见目标图重绘；微小像素差源于PDF渲染。作者姓名与固定PDF单位、公开Scholar单位相符。已逐张查看300dpi导出、确认完整。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Jianfei Cai；40906；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[312, 201, 565, 440]，PDF点、左上角原点。

Figure 1. Comparing generative open-ended object detection with
other open-set object detection tasks. Open-vocabulary object de-
tection and phrase grounding typically require predefined cate-
gories or phrases in text prompts to align with image regions. In
contrast, our introduced generative open-ended object detection is
a more general and practical setting where categorical information
is not explicitly defined. Such a setting is especially meaningful
for scenarios where users lack precise knowledge of object cate-
gories during inference.

最早核实视觉日期：2024-02-20；arXiv v1：2024-02-08；CVPR 2024 fixed official PDF

对照v1、v2、v3：v1框架右侧检测示例及教师/学生分支细节不同；固定会议图最早匹配v2（2024-02-20），不用v1日期。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Weidi Xie；25742；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[52, 70, 541, 222]，PDF点、左上角原点。

Figure 2. Illustration of the process for finetuning diffusion model and training the grounding head: (a) stable diffusion model is fine-tuned
on the detection dataset on base categories. (b) The grounding head is trained on synthetic images, with supervised learning on base
categories and self-training on novel categories.

最早核实视觉日期：2024-02-08；arXiv v1：2024-02-08；CVPR 2024 fixed official PDF

图1在v1版下移8pt；校正位置后视觉元素、图内文字与结果一致，微小偏移/压缩差异不构成重绘。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Weidi Xie；25742；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[51, 190, 545, 314]，PDF点、左上角原点。

Figure 1. (a) The synthetic images generated from Stable Diffusion and our proposed InstaGen, which can serve as a dataset synthesizer
for sourcing photo-realistic images and instance bounding boxes at scale. (b) On open-vocabulary detection, training on synthetic images
demonstrates significant improvement over CLIP-based methods on novel categories. (c) Training on the synthetic images generated from
InstaGen also enhances the detection performance in close-set scenario, particularly in data-sparse circumstances.

最早核实视觉日期：2025-01-31；arXiv v1：2025-01-31；CVPR 2025 fixed official PDF

已逐页比较arXiv v1与会议图4；整体图缩放与位置不同，节点、连接、示例图、配色和图例一致，作为同一视觉内容的排版缩放记录。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Wei-Shi Zheng；31913；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[71, 72, 542, 280]，PDF点、左上角原点。

Figure 4. The overview of LLMDet. LLMDet contains a standard open-vocabulary detector and a large language model (LLM) and is
trained under both grounding loss and language modeling loss. The LLM is designed to generate both image-level captions using feature
maps as visual input and region-level captions using a single object query as visual input, which are separated by different prompts.
Only vision tokens in region-level generation pass through the cross-attention (CA) modules in LLM, which is highlighted by a dashed
boundary. Since the number of tokens in image-level and region-level generation varies greatly, we forward the LLM twice separately to
save memory and computation. The LLM can be discarded in the inference time so that there is no extra cost.

最早核实视觉日期：2025-01-31；arXiv v1：2025-01-31；CVPR 2025 fixed official PDF

已查看固定PDF及目标图；以最早已核实匹配版本日期作时间代理，不声称实际作图日期。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Wei-Shi Zheng；31913；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[58, 66, 555, 271]，PDF点、左上角原点。

Figure 2. An example of GroundingCap-1M. Bounding box annotations are discarded for clarity. Compared with original short grounding
texts, the detailed captions in GroundingCap-1M are rich in object types, textures, colors, parts of the objects, object actions, precise object
locations and texts. Each caption in GroundingCap-1M has around 115 words on average.

最早核实视觉日期：2025-03-29；arXiv v1：2025-03-29；CVPR 2025 fixed official PDF

唯一arXiv v1于2025-03-29公开。固定会议版与v1裁图对照保存；已读图注、引言和第3节整体方法。论文Toronto身份与Steven L. Waslander公开Scholar18395次匹配。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Steven L. Waslander；18395；快照2026-10-22。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[115, 72, 500, 314]，PDF点、左上角原点。

Figure 2. Diagram of our proposed method. Offline Labeller Training: we add a detector head to a frozen DINOv2 encoder and train
it with source images only. Offline Label Generation: we combine and freeze the labeller backbone and detector, and generate target
pseudo-labels. Online Student Training: We train a student network using source ground truth boxes and target pseudo-labels, and align
patch features to a frozen DINOv2 encoder. During inference, the alignment encoder and projection MLP are not used.

最早核实视觉日期：2025-03-29；arXiv v1：2025-03-29；CVPR 2025 fixed official PDF

唯一arXiv v1于2025-03-29公开。固定会议版与v1裁图对照保存；已读图注、引言和第3节整体方法。论文Toronto身份与Steven L. Waslander公开Scholar18395次匹配。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Steven L. Waslander；18395；快照2026-10-22。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[322, 234, 548, 443]，PDF点、左上角原点。

Figure 1. Cosine similarity of patch feature from Cityscapes to
BDD100k.. We evaluate the similarity between the yellow star re-
gion in the Cityscapes image and all other regions in both images.
Compared to the EMA Teacher, DINOv2 generates semantically
stable features across domains, justifying the choice to use it for
feature alignment and as a frozen backbone for our labeller.

最早核实视觉日期：2025-07-03；arXiv v1：2025-07-03；arXiv v1 fixed

固定采集2025-07-03 arXiv v1图2；后来v2/v3和CVPR2026不用于追溯该样本。已查看v1整体框架和第III节；未发现更早投稿证据。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Xue Yang；14880；快照2026-10-22。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[52, 52, 561, 353]，PDF点、左上角原点。

Fig. 2. The overview of the proposed PWOOD framework. Orientation Learning and Scale Learning modules enables the Orientation-and-Scale-aware Student
to learn both scale and orientation information from weakly annotated data, as well as Class-Agnostic Pseudo-Label Filtering mechanism enhances pseudo-label
quality by leveraging dynamic thresholds.

最早核实视觉日期：2025-07-03；arXiv v1：2025-07-03；arXiv v1 fixed

固定采集2025-07-03 arXiv v1图1；后来v2/v3和CVPR2026不用于追溯该样本。已查看v1完整图与引言；未发现更早投稿证据。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Xue Yang；14880；快照2026-10-22。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[311, 193, 565, 497]，PDF点、左上角原点。

Fig. 1. The main paradigmatic types of existing oriented object detec-
tion. Existing methods are primarily divided into rotated boxes (RBox)-
supervised, horizontal boxes (HBox)-supervised, Point-supervised, and Semi-
Supervised approaches. Compared to these settings, our proposed Partial
Weakly-supervised methods have high annotation speed and low costs.

最早核实视觉日期：2023-11-23；arXiv v1：2023-11-23；CVPR 2024 fixed official PDF

PointOBB图1在v1上移10pt，校正历史裁剪坐标后图1和图2均像素完全一致；门槛作者改用Junchi Yan，固定PDF的SJTU与Scholar任职经历SJTU(18-)相符。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Junchi Yan；47501；快照2026-10-22。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[56, 74, 539, 358]，PDF点、左上角原点。

Figure 2. The pipeline of the PointOBB. PointOBB contains three views in total. From these views, an angle acquisition module and
a scale augmentation module are constructed. The scale augmentation module incorporates a Scale-Sensitive Consistency (SSC) loss to
enhance scale perception ability. Within the angle acquisition module, a Dense-to-Dense sample assignment is designed for angle learning,
optimized using the Self-Supervised Angle (SSA) loss. Additionally, a Dense-to-Sparse (DS) matching strategy is proposed to obtain more
precise object angles. During the training, a progressive multi-view switching strategy is designed to switch the resized and rot/flp views,
along with their corresponding modules.

最早核实视觉日期：2023-11-23；arXiv v1：2023-11-23；CVPR 2024 fixed official PDF

PointOBB图1在v1上移10pt，校正历史裁剪坐标后图1和图2均像素完全一致；门槛作者改用Junchi Yan，固定PDF的SJTU与Scholar任职经历SJTU(18-)相符。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Junchi Yan；47501；快照2026-10-22。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[308, 250, 547, 516]，PDF点、左上角原点。

Figure 1. The main paradigmatic types of existing oriented ob-
ject detection. Existing methods are primarily divided into rotated
boxes (RBox)-supervised, horizontal boxes (HBox)-supervised,
and Point-supervised approaches. Compared to RBox and HBox
labels, point labels have lower costs and higher efficiency.

最早核实视觉日期：2025-06；arXiv v1：2024-09-26；CVPR 2025 fixed official PDF

已对照唯一arXiv v1及会议版：全局流程相似，但v1的Heatmap Generation分支中蓝色框为空，会议版加入DINOv2标签，并重新压缩图像。不能视为完全相同视觉版本。以CVF官方BibTeX的June 2025记录会议固定图的最早可核实公开月份，整月均在S2。历史PDF仅用于差异对照。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Heng Ji；43364；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[57, 70, 555, 424]，PDF点、左上角原点。

Figure 2. The entire architecture of our method. We compare the adjusted embeddings, produced by the DINOv2 model, of the positive
and negative support images, with the relevant masks extracted using the SAM model to provide an initial estimate of our segmentation
BBox. We again use DINOv2 for generating pixel-precise heatmaps which provide another estimate for the segmentation. We combine
both these estimates using a binarized overlap to get the final segmentation mask.

最早核实视觉日期：2025-06；arXiv v1：2024-09-26；CVPR 2025 fixed official PDF

已对照唯一arXiv v1（2024-09-26）：图1下排为Dog/Aerial Boat；会议版换为Phone/Fork并调整尺度，不能回填v1日期。固定会议图的可核实公开时间为CVF官方BibTeX注明的June 2025，保留月份精度及整月范围；该范围完全位于S2。历史PDF仅作非匹配版本对照。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Heng Ji；43364；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[316, 230, 547, 443]，PDF点、左上角原点。

Figure 1. Detection results for label “Mountain Dew”. While the
GroundingDINO, one of the state-of-the-art grounded object de-
tection methods, fails to capture the Mountain Dew bottles in the
image displayed in the figure, SearchDet manages to ground every
instance of Mountain Dew that appears in the image. Detection
results for other classes “Phone” and “Fork”.

最早核实视觉日期：2025-03-28；arXiv v1：2025-03-28；CVPR 2025 fixed official PDF

已查看唯一arXiv v1(2025-03-28)和固定会议版目标图；构图、元素、连线、文字相同。PDF对象证据显示v1使用DeviceRGB，会议版使用DeviceCMYK/Indexed-CMYK，黑底及照片有出版色彩转换和压缩差异；不将此机械转换认定为重绘。保留两版完整裁图以供复核。作者Tianjin University身份匹配Wei Feng公开Scholar11058次。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Wei Feng；11058；快照2026-10-22。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[58, 61, 554, 206]，PDF点、左上角原点。

Figure 4. The detailed architecture of our method. Initially, we sample instances from the old data according to their classes. Subsequently,
during the class combination, we identify the potential old classes through the old model. In the instance selection, we select the instances
of potential classes from the instance pool. After that, we calculate the position in new image to replay the instance and fuse them with the
new image to create a fused image. Finally, the fused image is trained in an enhanced way.

最早核实视觉日期：2025-03-28；arXiv v1：2025-03-28；CVPR 2025 fixed official PDF

已查看唯一arXiv v1(2025-03-28)和固定会议版目标图；构图、元素、连线、文字相同。PDF对象证据显示v1使用DeviceRGB，会议版使用DeviceCMYK/Indexed-CMYK，黑底及照片有出版色彩转换和压缩差异；不将此机械转换认定为重绘。保留两版完整裁图以供复核。作者Tianjin University身份匹配Wei Feng公开Scholar11058次。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Wei Feng；11058；快照2026-10-22。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[318, 263, 554, 436]，PDF点、左上角原点。

Figure 1. Comparison of traditional image replay (a) and our re-
play methods (b). (a) shows that only the old class “horse” is la-
beled in stored image, while other classes (new class “person” and
old class “car”) are labeled as background. Both old (“horse”) and
future classes in new images are labeled as background. (b) Our
method avoids confusing information in stored image by retaining
instances and alleviates background shift by fusing these instances
in the new image.

最早核实视觉日期：2025-04-07；arXiv v1：2025-04-07；CVPR 2025 fixed official PDF

唯一arXiv版本v1与固定CVPR图同坐标300dpi裁剪完全一致（MAE=0）；作者Ming-Ming Cheng的南开大学身份与引用主页一致。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Ming-Ming Cheng；85763；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[60, 70, 552, 274]，PDF点、左上角原点。

Figure 4. Illustration of our DFormerv2. (a) Overall architecture of our DFormerv2, which contains an encoder with pyramid structure
and a decoder head that receives input from the last three stage features. (b) Detailed structure of the basic building block. (c) Detailed
illustration of the proposed geometry self-attention mechanism.

最早核实视觉日期：2025-04-07；arXiv v1：2025-04-07；CVPR 2025 fixed official PDF

唯一arXiv版本v1与固定CVPR图同坐标300dpi裁剪完全一致（MAE=0）；作者Ming-Ming Cheng的南开大学身份与引用主页一致。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Ming-Ming Cheng；85763；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[333, 212, 535, 422]，PDF点、左上角原点。

Figure 1. Comparisons among the main RGBD segmentation
pipelines and our approach. (a) Use dual encoders to encode
RGB and depth respectively and design fusion modules to fusion
them [28, 61]; (b) Adopt an unified RGBD encoder to extract and
fuseRGBDfeatures[1,59]; (c)OurDFormerv2usedepthtoform
ageometrypriorofthesceneandthenenhancethevisualfeatures.

最早核实视觉日期：2025-05-16；arXiv v1：2025-05-16；CVPR 2025 fixed official PDF

已查看唯一v1（2025-05-16）与会议版，完整框架一致。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Song Wang；16213；快照2026-10-22。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[56, 68, 557, 277]，PDF点、左上角原点。

Figure 4. Architecture of our DPSeg network. We begin by generating visual prompts based on text prompt templates, which are then
combined with text prompts to create the dual-prompt cost volume. Subsequently, we incorporate multi-scale image features and corre-
sponding visual prompt embeddings into the cost volume-guided decoder (CVGD) in a progressive manner.

最早核实视觉日期：2025-05-16；arXiv v1：2025-05-16；CVPR 2025 fixed official PDF

已查看唯一v1（2025-05-16）与会议版，目标图构图和标签一致。右端坐标刻度在原PDF嵌入图中已不完整，扩大裁剪仍相同，保留原图，不补字。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Song Wang；16213；快照2026-10-22。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[315, 230, 560, 339]，PDF点、左上角原点。

Figure 1. T-SNE visualization of embeddings in CLIP feature
space, showing text prompt, visual prompt, target image, and
our dual-prompt for a street scene image. Numbers indicate dis-
tance to image feature, where our dual-prompt approach achieves
closest proximity (0.18).

最早核实视觉日期：2025-03-16；arXiv v1：2024-11-26；CVPR 2025 fixed official PDF

v1图1尚使用Embedding文字且存在底纹等差异；v2完整图匹配会议版，逐像素一致，v3未改变。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Ming-Hsuan Yang；189781；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[317, 245, 554, 505]，PDF点、左上角原点。

Figure 1. We present CASS, object-level Context-Aware
training-free open-vocabulary Semantic Segmentation model. (a)
Overview: Our method distills the vision foundation model’s
(VFM)object-levelcontextualspectralgraphintoCLIP’sattention
and refines query text embeddings towards object-specific seman-
tics. (b) Object-Level Context: This result illustrates how incor-
porating object-level context improves segmentation accuracy by
unifying object-wise components into a user-defined object class;
forexample,(toprow)ourapproachpreciselysegmentsthetruck’s
body, wheels, and cargo area and (bottom row) accurately groups
elements into a single object, such as the sheep, whereas base-
lines [16, 29] often fail to achieve this unified segmentation.

最早核实视觉日期：2024-11-26；arXiv v1：2024-11-26；CVPR 2025 fixed official PDF

v1图3与会议图逐像素一致。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Ming-Hsuan Yang；189781；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[58, 71, 294, 262]，PDF点、左上角原点。

Figure 3. Attention score visualization for various query points.
Left: Vanilla CLIP (A ) shows noisy, unfocused attention.
CLIP
Center: VFM-to-CLIP distillation without low-rank eigenscaling
shows partial object grouping with limited detail. Right: Incor-
porating our low-rank eigenscaling (Eq. (7)) captures object-level
context, improving grouping within a single object.

最早核实视觉日期：2024-06-17；arXiv v1：2024-06-17；CVPR 2024 fixed official PDF

已读源PDF图注与正文并查看图页；作者单位核验一致。已逐张查看300dpi完整导出；对应arXiv v1视觉像素一致。实际作图日期未知。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Yunchao Wei；37025；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[50, 72, 547, 344]，PDF点、左上角原点。

Figure 2. Framework of our WeCLIP. The image is input to the Frozen CLIP image encoder to generate the image features, and class
labels are used to build text prompts and then input to the Frozen CLIP text encoder to generate the text features. The classification scores
are generated based on the distance between the pooled image and text features. Using GradCAM, we can generate the initial CAM
M . Then, the frozen image features from the last layer of each transformer block are input to our decoder to generate the final semantic
init
segmentation predictions. Meanwhile, the affinity map A from our decoder and the multi-head attention maps A from CLIP are input to
f s
our RFM to establish refining maps R to refine M as M . After post-processing, it will be used as the supervision to train our decoder.
init f

最早核实视觉日期：2024-06-17；arXiv v1：2024-06-17；CVPR 2024 fixed official PDF

已读源PDF图注与正文并查看图页；作者单位核验一致。已逐张查看300dpi完整导出；对应arXiv v1视觉像素一致。实际作图日期未知。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Yunchao Wei；37025；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[308, 230, 536, 358]，PDF点、左上角原点。

Figure 1. Comparisons between our approach and other single-
stage or CLIP-based approaches. (a) Previous single-stage ap-
proach, which uses a trainable ImageNet [11] pre-trained back-
bone with trainable classification and segmentation process. (b)
Previous CLIP-based approach, which is a multi-stage approach
that uses the Frozen CLIP model to produce pseudo labels and
trains an individual ImageNet pre-trained segmentation model. (c)
Our approach. Our approach is a single-stage approach that uses
a frozen CLIP model as the backbone with a trainable segmenta-
tion process, significantly reducing the training cost.

最早核实视觉日期：2024-03-12；arXiv v1：2024-03-12；CVPR 2024 fixed official PDF

已读源PDF图注与正文并查看图页；作者单位核验一致。已逐张查看300dpi完整导出；对应arXiv v1视觉像素一致。实际作图日期未知。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Zongyuan Ge；23901；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[53, 72, 539, 286]，PDF点、左上角原点。

Figure 2. Overview of the proposed unified learning framework. (a) shows image label-guided WSSS (from classification to segmentation).
The upper branch describes the classification network θ identifying object regions corresponding to each category to minimize LBC E .
Introduce a self-supervised learning paradigm using context prototype-aware learning to provide a more complete CAM, supervising the
initial CAM and minimizing LS elf . The lower branch refines these CAMs (e.g., DenceCRF [26]) to form pseudo-labels for supervising the
semantic segmentation network. (b) outlines our strategy based on context prototype-aware learning. In mini-batches, instance prototypes
P I are generated using CAM and extracted features f , updating the support bank. Then, the bank is used to construct a context prototype
n
set P c . Feature distribution alignment is then applied to the current instance features, adding a shift term δ to guide them toward clusters of
n n
dense features in the bank. Next, soft neighbors are softly measured for P I based on P c , with P I serving as anchors. Finally, positiveness
n n n
value w can be computed between two specific attributes. This mechanism selects K soft positive neighbors P˜ c to generate PACAM.
i n

最早核实视觉日期：2024-03-12；arXiv v1：2024-03-12；CVPR 2024 fixed official PDF

已读源PDF图注与正文并查看图页；作者单位核验一致。已逐张查看300dpi完整导出；对应arXiv v1视觉像素一致。实际作图日期未知。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Zongyuan Ge；23901；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[313, 285, 540, 436]，PDF点、左上角原点。

Figure 1. The main idea promoted throughout the paper is that
semantic context prototype-aware underpins localization of indi-
vidual objects in WSSS. Our CPAL performs adaptive perception
of diverse attributes (e.g., cat) with attribute hunting (c) rather
than from single prototype (a) and plain context prototypes (b).
This attribute-specific adaptation not only mitigates the risk of er-
rors where (b) mistakenly identifies similar categories (e.g., dog)
but also ensures accurate activation of the complete object region.

最早核实视觉日期：2024-03-30；arXiv v1：2024-03-30；CVPR 2024 fixed official PDF

已读源PDF图注与正文并查看图页；作者单位核验一致。已逐张查看300dpi完整导出；对应arXiv v1视觉像素一致。实际作图日期未知。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Tianzhu Zhang；23149；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[60, 75, 534, 336]，PDF点、左上角原点。

Figure 2. Framework of our proposed Relation-aware Intra-modal Matching (RIM) Network. We first explore Stable Diffusion model and
SAM to construct image-modal reference features, then we conduct relation-aware matching between region features and reference features
based on ranking distribution. The matching is established in the all-purpose feature spaces of DINOv2.

最早核实视觉日期：2024-03-30；arXiv v1：2024-03-30；CVPR 2024 fixed official PDF

已读源PDF图注与正文并查看图页；作者单位核验一致。已逐张查看300dpi完整导出；对应arXiv v1视觉像素一致。实际作图日期未知。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Tianzhu Zhang；23149；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[314, 245, 545, 418]，PDF点、左上角原点。

Figure 1. Motivation of our method. (a) False matches tend to
occur in cross-modal features. We establish well-aligned image-
modal reference features thus transit the text-to-image matching to
image-to-image matching. (b) Indivdual matching tends to suffer
from disturbances. We propose a novel relation-aware matching
strategy for more robust region classification.

最早核实视觉日期：2025-03-24；arXiv v1：2025-03-24；CVPR 2025 fixed official PDF

已逐页查看会议版与arXiv v1；图区设计一致，保留移动/缩放及PDF压缩差异，不声称像素完全一致。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Lei Zhang；137335；快照2026-10-21。

绘图工具：unknown。已读图注与相关正文；未找到该插图实际制作工具声明。

裁剪坐标：[61, 70, 534, 315]，PDF点、左上角原点。

Figure 3. Architecture of MaSSFormer. The model architecture is shown on the top left corner, including the image encoder, pixel decoder
and transformer decoder. The detailed structure of the high-resolution pixel decoder is shown on the right of this figure.

最早核实视觉日期：2025-03-24；arXiv v1：2025-03-24；CVPR 2025 fixed official PDF

已逐页查看会议版与arXiv v1；图区设计一致，保留移动/缩放及PDF压缩差异，不声称像素完全一致。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Lei Zhang；137335；快照2026-10-21。

绘图工具：unknown。已读图注与相关正文；未找到该插图实际制作工具声明。

裁剪坐标：[73, 191, 522, 400]，PDF点、左上角原点。

Figure 1. Comparisons on the resolution and annotation quality between existing semantic segmentation datasets (a) COCO, (b) ADE20K,
(c) EntitySeg and (d) our established dataset MaSS13K. Note that the ‘others’ category, which is highlighted in yellow, in our MaSS13K
actually contains a group of well-segmented objects without specific class names. Please zoom-in for a better view.

最早核实视觉日期：2025-04-04；arXiv v1：2025-04-04；CVPR 2025 fixed official PDF

已查看v1、v2及会议PDF，整体图形一致；F的固定区域三版像素一致。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Robby T. Tan；18517；快照2026-10-21。

绘图工具：unknown。已读图注与相关正文；未找到该插图实际制作工具声明。

裁剪坐标：[68, 71, 545, 259]，PDF点、左上角原点。

Figure 2. Overall architecture of MFuser. MFuser takes inputs through both VFM and VLM visual encoders. Features from each encoder
layer are concatenated and refined in MVFuser, which captures sequential and spatial dependencies in parallel. The refined features are
then added back to the original features and passed to the next layer. MTEnhancer strengthens text embeddings of each class by integrating
visualfeaturesthroughahybridattention-Mambamechanism. TheenhancedtextembeddingsserveasobjectqueriesfortheMask2Former
decoder, alongside multi-scale visual features. During training, only MVFusers, MTEnhancers, and the segmentation decoder are trainable
while the VFM and VLM remain frozen, preserving their generalization ability and enabling efficient training. Note that skip connections
between each block of MTEnhancer are omitted for clarity.

最早核实视觉日期：2025-04-04；arXiv v1：2025-04-04；CVPR 2025 fixed official PDF

已查看v1、v2及会议PDF，整体图形一致；F的固定区域三版像素一致。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Robby T. Tan；18517；快照2026-10-21。

绘图工具：unknown。已读图注与相关正文；未找到该插图实际制作工具声明。

裁剪坐标：[322, 231, 549, 407]，PDF点、左上角原点。

Figure 1. Comparative analysis of the VFM and the VLM fea-
tures. VFM: Visualization of PCA-computed features from DI-
NOv2 (the first three components of PCA, computed on the image
features, serve as color channels), displaying fine-grained details
but lacking text alignment. VLM: Image-text similarity map from
EVA02-CLIPusingthequery‘car’,demonstratinggoodalignment
with text but insufficient localization of queried objects. MFuser:
OurproposedfusionframeworkintegratesVFMandVLM,result-
ing in unified features that exhibit both precise locality and robust
text alignment. Quantitative results on synthetic-to-real DGSS
benchmarks further validate our approach, with MFuser consis-
tently achieving the highest mIoU scores across all tasks.

最早核实视觉日期：2025-03-27；arXiv v1：2024-12-16；CVPR 2025 fixed official PDF

已查看v1、v2及会议版。F的LASS和归一化等模块连线/文字发生变化，会议图3匹配v2；不得套用v1日期。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Yizhou Yu；35607；快照2026-10-21。

绘图工具：unknown。已读图注与相关正文；未找到该插图实际制作工具声明。

裁剪坐标：[76, 71, 536, 322]，PDF点、左上角原点。

Figure 3. Overall Architecture of SegMAN. (a) Hierarchical SegMAN Encoder. (b) LASS for modeling global contexts and local details
with linear complexity. (c) The SegMAN Decoder. (d) The MMSCopE module for multi-scale contexts extraction.

最早核实视觉日期：2024-12-16；arXiv v1：2024-12-16；CVPR 2025 fixed official PDF

已查看v1、v2及会议版。M在v1为图1，会议版改号图2并等比例缩放；问题图的内容和设计一致。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Yizhou Yu；35607；快照2026-10-21。

绘图工具：unknown。已读图注与相关正文；未找到该插图实际制作工具声明。

裁剪坐标：[87, 71, 523, 308]，PDF点、左上角原点。

Figure 2. Qualitative analysis of receptive field patterns and segmentation performance for small-sized models (27M-29M parameters).
Left: Visualization of effective receptive fields (ERF) on the Cityscapes validation set at 1024×1024 resolution, illustrating SegMAN’s
stronger global context modeling capacity in comparison to existing state-of-the-art models. Right: Segmentation maps highlighting
SegMAN’s superior capacity to encode fine-grained local details that are often missed by existing approaches.

最早核实视觉日期：2024-11-20；arXiv v1：2024-11-20；CVPR 2025 fixed official PDF

已阅读固定版本图注和正文；与arXiv历史对应视觉比较，具体差异记入逐图核查记录。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Lei Zhang；137335；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[58, 65, 590, 197]，PDF点、左上角原点。

Figure3. IllustrationofthetrainingandinferenceprocessesofAdcSR,aninstantiationofourADCframeworkappliedtoOSEDiff.
(a)InStage1,wepretrainaprunedVAEdecoderthatsharesthelatentspacewithSDandOSEDiff. (b)InStage2,wedistilltheknowledge
from OSEDiff (ADC-teacher) into AdcSR (ADC-student) by aligning features in the pretrained decoder. An adversarial loss encourages
the student to generate features that can fool a LoRA-finetuned SD UNet (ADC-discriminator), utilizing the corresponding real features
of GT images. Since all supervisions perform in the feature space, there is no need to decode images as in previous approaches [96, 112].
(c) During inference, the LR image is directly fed into our trained compressed UNet and VAE decoder to obtain the super-resolution result.

最早核实视觉日期：2024-11-20；arXiv v1：2024-11-20；CVPR 2025 fixed official PDF

v1与会议图的构图、标签、颜色与箭头匹配，仅整体纵向排版移动19.925pt；采用v1日期。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Lei Zhang；137335；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[317, 260, 555, 401.5]，PDF点、左上角原点。

Figure 1. Comparison between our proposed AdcSR and typ-
ical one-step diffusion-based Real-ISR methods. (a) The state-
of-the-art one-step diffusion network OSEDiff [96] employs com-
plete SD [80] models for Real-ISR, suffering from high computa-
tional costs. (b) We distill OSEDiff (ADC-teacher) into a smaller
diffusion-GAN hybrid model, AdcSR (ADC-student), under the
proposed ADC framework, achieving significantly improved effi-
ciency while maintaining competitive recovery performance.

最早核实视觉日期：2025-03-13；arXiv v1：2024-11-27；CVPR 2025 fixed official PDF

已逐项比对v1、v2和会议版：v1专家模块结构和文字不同；v2(2025-03-13)构图、配色、连线及标签与会议版一致，不套用v1日期。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Yulun Zhang；40552；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[57, 70, 555, 257]，PDF点、左上角原点。

Figure 2. Proposed MoCE-IR framework. Despite recent advances in MoE-based image restoration [57, 62, 63], inconsistent expert
behavior—where some experts over-generalize while others underperform—limits their computational efficiency. We address this through
complexity experts: flexible blocks with varying computational capacity and receptive fields. Our MoCE-IR employs an asymmetric
encoder-decoder architecture where each decoder block contains a mixture-of-complexity-experts layer for adaptive capacity routing.

最早核实视觉日期：2024-11-27；arXiv v1：2024-11-27；CVPR 2025 fixed official PDF

已查看固定PDF及目标图；以最早已核实匹配版本日期作时间代理，不声称实际作图日期。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Yulun Zhang；40552；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[316, 225, 555, 416]，PDF点、左上角原点。

Figure 1. Motivation. (a) Dense all-in-one restoration meth-
ods [24, 57] often inefficiently allocate parameters when handling
multiple degradation types. (b) While recent Mixture-of-Experts
(MoE) approaches [61, 62] address this through sparse computa-
tion, their rigid routing mechanisms uniformly distribute inputs
across experts without considering the natural relationships be-
tween degradations. (c) To overcome these limitations, we intro-
duce Complexity Experts - adaptive processing blocks with size-
varying computational units. Our framework dynamically allocates
model capacity using a spring-inspired force mechanism that con-
tinuously guides routing decisions toward simpler experts when
possible, with the force proportional to the complexity of the input
degradation. While initially designed for computational efficiency,
this approach naturally emerges as a task-discriminative learning
framework, assigning degradations to the most suitable experts.
This makes it particularly effective for all-in-one restoration meth-
ods, where both task-specific processing and cross-degradation
knowledge sharing are crucial.

最早核实视觉日期：2025-03-19；arXiv v1：2024-10-15；CVPR 2025 fixed official PDF

逐一查看v1(2024-10-15)、v2(10-30)、v3(11-06)、v4(2025-03-19)。v1底色及模块布局不同；v2/v3采用新布局但残差块配色等细节仍不同。v4与固定会议图同坐标300dpi裁剪像素完全一致（MAE=0），故使用v4日期。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Jinshan Pan；26700；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[80, 73, 532, 284]，PDF点、左上角原点。

Figure 3. Overview of DORNet. Given D as input, the degradation learning first encodes it to produce degradation representations D˜
up
and D. Then, D˜, D, D , and I are fed into multiple degradation-oriented feature transformation (DOFT) modules, generating the HR
lr
depth D . Finally, D and D are sent to the degradation regularization to obtain D , which is used as input for the degradation loss
hr hr d
L and the contrastive loss L . The degradation regularization only applies during training and adds no extra overhead in testing.
deg cont

最早核实视觉日期：2025-03-19；arXiv v1：2024-10-15；CVPR 2025 fixed official PDF

逐一查看v1(2024-10-15)、v2(10-30)、v3(11-06)、v4(2025-03-19)。v1/v2缺少底部退化正则回路；v3增加回路但标签/文字视觉仍不同。v4与固定会议图同坐标300dpi裁剪像素完全一致（MAE=0），故使用v4日期。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Jinshan Pan；26700；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[317, 244, 552, 417]，PDF点、左上角原点。

Figure 1. Previous methods (a) directly fuse the RGB information
aligned with the LR depth, while our method (b) focuses more on
modeling the degradation representation of the LR depth to pro-
vide targeted guidance for HR depth recovery.

最早核实视觉日期：2025-06；arXiv v1：2024-11-27；CVPR 2025 fixed official PDF

仅有arXiv v1（2024-11-27），已查看。会议图与v1不完全相同：框架重绘，改变模块位置、训练噪声分支和图例排布。不把会议视觉版本回溯到v1。使用CVF官方BibTeX明确的2025年6月出版月份作为当前最早已核实证据；精度为月，不声称首次绘制日。历史PDF及图仅供差异对照。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Jinshan Pan；26700；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[65, 72, 548, 226]，PDF点、左上角原点。

Figure 2. An overview of the FaithDiff, which takes LQ images and image descriptions as inputs and restores HQ images via diffusion
process. To fully leverage the power of LDMs, we propose to unleash diffusion priors. An alignment module is developed to effectively
incorporate the features extracted from LQ images with the noisy latent of the diffusion model. We jointly optimize the encoder, the
alignment module, and the diffusion model, which can benefit from their interplay and lead to faithful SR images with high visual quality.

最早核实视觉日期：2025-06；arXiv v1：2024-11-27；CVPR 2025 fixed official PDF

仅有arXiv v1（2024-11-27），已查看。会议图与v1不完全相同：图内SeeSR/SUPIR引文编号从41/44更新为44/49；其他构图相同。不把会议视觉版本回溯到v1。使用CVF官方BibTeX明确的2025年6月出版月份作为当前最早已核实证据；精度为月，不声称首次绘制日。历史PDF及图仅供差异对照。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Jinshan Pan；26700；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[59, 70, 551, 221]，PDF点、左上角原点。

Figure 1. Visual comparison with state-of-the-art SR methods. (b) and (c) are the intermediate results of the degradation removal module
(DRM) by DiffBIR [21] and SUPIR [49]. The competing methods [21, 49] do not effectively restore faithful structures based on the
degradation removal results in (b)-(c). In contrast to the results in (d) and (f)-(g), our approach can recover more realistic high-quality
results with faithful contents (e.g., the characters in (h)).

最早核实视觉日期：2025-03-17；arXiv v1：2025-03-17；CVPR 2025 fixed official PDF

核对arXiv v1（2025-03-17）和CVPR固定版本：布局、配色、连线、子图和文字匹配；栅格压缩/渲染有轻微像素差异。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Ying Tai；19268；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[70, 72, 535, 273]，PDF点、左上角原点。

Figure 3. Overall framework of our proposed ERR. Building on the insight from progressive spectral perspective, ERR consists of three
collaborative sub-networks: the zero-frequency enhancer (ZFE), the low-frequency restorer (LFR), and the high-frequency refiner (HFR).

最早核实视觉日期：2025-03-17；arXiv v1：2025-03-17；CVPR 2025 fixed official PDF

核对arXiv v1（2025-03-17）和CVPR固定版本：布局、配色、连线、子图和文字匹配；栅格压缩/渲染有轻微像素差异。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Ying Tai；19268；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[70, 70, 541, 357]，PDF点、左上角原点。

Figure 2. Our core motivation. Based on the observations in (a) and (b), we deconstruct the complex UHD restoration problem into three
progressive stages: zero-frequency enhancement, low-frequency restoration, and high-frequency refinement.

最早核实视觉日期：2024-11-26；arXiv v1：2024-11-26；CVPR 2025 fixed official PDF

已阅读固定版本图注和正文；与arXiv历史对应视觉比较，具体差异记入逐图核查记录。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Vishal M. Patel；48975；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[82, 70, 530, 283]，PDF点、左上角原点。

Figure 4. (a) Illustrates the training stage of the GenDeg model where it is trained to condition on the clean image, text prompt and mean
intensity (µ) and variation (σ) of the degradation pattern. (b) Shows the inference stage where the model generates a degraded image based
on these conditions; and (c) Depicts the architecture of the Swin-transformer-based restoration network.

最早核实视觉日期：2025-03-22；arXiv v1：2024-11-26；CVPR 2025 fixed official PDF

v1图2与会议图2的柱高及百分比标注不同，如Haze +1619/+139改为+1503/+135；v2图2逐像素匹配会议图，采用2025-03-22，不回填v1日期。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Vishal M. Patel；48975；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[318, 72, 554, 239]，PDF点、左上角原点。

Figure 2. Analysis of real and synthetic image restoration datasets
for various degradations. Existing datasets are small and less di-
verse, especially for haze, low-light, and raindrop. Our diffusion-
generated synthetic data substantially increases the number of
samples as well as scene diversity.

最早核实视觉日期：2024-11-26；arXiv v1：2024-11-26；arXiv v1 fixed source

固定arXiv v1原件；会议视觉与早期版本存在差异，未给新版图套用v1日期。TARS已查看v1/v2及会议版；PassionSR已查看v1/v2/v3及会议版；具体差异保留在review页图中。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Yulun Zhang；40552；快照2026-10-21。

绘图工具：unknown。已读图注与相关正文；未找到该插图实际制作工具声明。

裁剪坐标：[58, 70, 554, 372]，PDF点、左上角原点。

Figure 4. Overview of our PassionSR. Step 1: we simplify OSEDiff [42] by removing DAPE and CLIP Encoder, obtaining PassionSR-
FP. Step 2: the quantizer we use has two key trainable parts, consisting of the Learnable Boundary Quantizer and Learnable Equivalent
Transformation. Step 3: we design a distributed calibration strategy and special loss function to accelerate convergence of calibration.

最早核实视觉日期：2024-11-26；arXiv v1：2024-11-26；arXiv v1 fixed source

固定arXiv v1原件；会议视觉与早期版本存在差异，未给新版图套用v1日期。TARS已查看v1/v2及会议版；PassionSR已查看v1/v2/v3及会议版；具体差异保留在review页图中。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Yulun Zhang；40552；快照2026-10-21。

绘图工具：unknown。已读图注与相关正文；未找到该插图实际制作工具声明。

裁剪坐标：[318, 70, 551, 150]，PDF点、左上角原点。

Figure 2. Visual comparison (×4) of one-step diffusion SR models. We use OSEDiff as a 32-bit full-precision (FP) reference and provide 6-bit quantized version with different methods.

最早核实视觉日期：2025-03-10；arXiv v1：2024-08-16；CVPR 2025 fixed official PDF

已查看v1（2024-08-16）和v2（2025-03-10）。v1整体图右侧含MSFM分图，AdaUp及查询模块布局亦不同。v2与会议目标图同坐标300dpi裁剪完全一致（MAE=0）；不沿用v1日期。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Zheng-Jun Zha；30479；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[62, 73, 555, 314]，PDF点、左上角原点。

Figure 2. The overall framework of our proposed QMambaBSR, primarily includes the novel Query State Space Model (QSSM), Multi-
scale Fusion Module (MSFM), and the Adaptive Up-sampling Module (AdaUp).

最早核实视觉日期：2025-03-10；arXiv v1：2024-08-16；CVPR 2025 fixed official PDF

已查看v1（2024-08-16）和v2（2025-03-10）。v1图1在第2页，强度曲线的数字/标记与目标图不同。v2与会议目标图同坐标300dpi裁剪完全一致（MAE=0）；不沿用v1日期。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Zheng-Jun Zha；30479；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[317, 213, 552, 414]，PDF点、左上角原点。

Figure 1. The concat and frame-by-frame operations in exist-
ing methods struggle to efficiently extract sub-pixels and suppress
noise, leading to remaining artifacts and over-smoothed details, as
shown in (a). We observe that noise randomly appears on several
frames, while effective sub-pixels have consistent intensity at cor-
responding positions in all frames, as shown in (c). Based on this,
a novel inter-frame query and intra-frame scanning-based QMam-
baBSR is proposed to extract more accurate sub-pixels while mit-
igating noise interference simultaneously, as shown in (b).

最早核实视觉日期：2025-05-26；arXiv v1：2025-05-26；CVPR 2025 fixed official PDF

arXiv v1（2025-05-26）为唯一历史版本，与会议目标图同坐标300dpi裁剪像素完全相同（MAE=0）。作者Deyu Meng的西安交大身份与公开引用主页一致。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Deyu Meng；49595；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[58, 71, 555, 255]，PDF点、左上角原点。

Figure3. IllustrationsofourproposedadaptivenetworkAdaReNet. Specifically, I ∈ RH×W×C representsanoisyimage, whereH andW
represent the spatial dimensions, and C denotes the channel dimension. The Vanilla Module and EQ Module each produce their respective
preliminary denoising results, denoted as f and f . The Fusion Module Mask(·) automatically decides which areas of the image to use
c e
moreEQModulewouldgainmorebenefit. AfteradaptivefusionbyMask(·)andcorrectionbytheSelf-correctingModuleS (·), thefinal
c
denoised image I¯is output.

最早核实视觉日期：2025-05-26；arXiv v1：2025-05-26；CVPR 2025 fixed official PDF

arXiv v1（2025-05-26）为唯一历史版本，与会议目标图同坐标300dpi裁剪像素完全相同（MAE=0）。作者Deyu Meng的西安交大身份与公开引用主页一致。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Deyu Meng；49595；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[319, 229, 553, 317]，PDF点、左上角原点。

Figure 1. Illustration of the output feature map of a typical image
obtained by standard CNN and our used rotation equivariant con-
volution neural network. Both networks are initialized randomly.

最早核实视觉日期：2025-03-18；arXiv v1：2025-03-18；CVPR 2025 fixed official PDF

已查看固定PDF及目标图；以最早已核实匹配版本日期作时间代理，不声称实际作图日期。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Vishal M. Patel；48975；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[58, 70, 555, 230]，PDF点、左上角原点。

Figure 3. This diagram illustrates our multimodal super-resolution pipeline. Starting with a low-resolution (LR) image, we extract modalities
like depth and semantic segmentation maps. These modalities are encoded into tokens and transformed into multimodal latent tokens (m).
Our diffusion model uses these tokens and the LR input to generate a high-resolution (SR) output. A multimodal classifier-free guidance
(m-cfg) refines the SR image for enhanced quality.

最早核实视觉日期：2025-03-18；arXiv v1：2025-03-18；CVPR 2025 fixed official PDF

已查看固定PDF及目标图；以最早已核实匹配版本日期作时间代理，不声称实际作图日期。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Vishal M. Patel；48975；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[316, 70, 555, 150]，PDF点、左上角原点。

Figure 2. Language models struggle to accurately represent spa-
tial information, leading to coarse and imprecise image super-
resolution. To overcome this limitation, we incorporate additional
spatialmodalitieslikedepthmapsandsemanticsegmentationmaps.
These modalities provide detailed spatial context, allowing our
model to implicitly align language descriptions with individual pix-
els through a transformer network. This enriched understanding of
theimagesignificantlyenhancestherealismofoursuper-resolution
results and minimizes distortion.

最早核实视觉日期：2024-07-01；arXiv v1：2024-07-01；CVPR 2025 fixed official PDF

已打开v1第3页及v2第3页。v1图1整体随版心均匀缩放；合成点云、UDF预测、网格提取及所有文字、连线一致。按对应图元坐标反算历史裁剪；最早匹配v1(2024-07-01)。图像压缩与缩放差异不视为重绘。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Junhui Hou；21404；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[58, 73, 555, 184]，PDF点、左上角原点。

Figure 1. Pipeline. First, we train a UDF prediction network U on a synthetic dataset, which contains a series of local point cloud patches
Θ
that are independent of specific shapes. Given a global point cloud P, we then extract a local patch P assigned to each query point q within
a specified radius, and obtain the corresponding UDF values U (P , q). Finally, we extract the mesh corresponding to the input point cloud
Θˆ
by incorporating the DCUDF [18] framework.

最早核实视觉日期：2025-05-10；arXiv v1：2024-07-01；CVPR 2025 fixed official PDF

核查v1(2024-07-01)与v2(2025-05-10)：v1图2位于第4页、图内分组标签相对几何示意的比例不同；固定会议图的分组文字排布首次在v2第3页匹配。几何内容未改变；以v2视觉排布为保守公开代理，仍在S2。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Junhui Hou；21404；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[58, 590, 297, 678.5]，PDF点、左上角原点。

Figure 2. Local geometries. (a) For points on a geometry that are
differentiable, the local shape at these points can be approximated
by quadratic surfaces. (b) For points that are non-differentiable, we
can also construct locally approximated surfaces using functions.

最早核实视觉日期：2023-12-21；arXiv v1：2023-12-21；CVPR 2024 fixed official PDF

会议F2与v1 p3图形完全匹配（MAE0）。会议M1与v2 p2完全匹配（MAE0）；v1 p2缺少MV/TP/RM行标签，且子图排版不同，故M不回溯v1。已实际查看两历史PDF、会议版图注与正文及导出。Hao Tan与Adobe作者身份一致；真实绘图日期未知。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Hao Tan；12273；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[64, 72, 534, 237]，PDF点、左上角原点。

Figure 2. Overview of Carve3D. Given a prompt sampled from our curated prompt set and a initial noisy image, we iteratively denoise
the image using the UNet. The final, clean image contains four multi-view images tiled in a 2-by-2 grid. MRC reward is computed by
comparing (a) the generated multi-view images with (c) the corresponding multi-view images rendered at the same camera viewpoints
from (b) the reconstructed NeRF. Then, we train the model with policy gradient loss function, where the loss is derived from the reward
and log probabilities of the UNet’s predictions, accumulated over all denoising timesteps. By using only a set of training text prompts, our
RLFT algorithm finetunes the diffusion model by evaluating its own generated outputs, without relying on ground truth multi-view images.

最早核实视觉日期：2024-04-09；arXiv v1：2023-12-21；CVPR 2024 fixed official PDF

会议F2与v1 p3图形完全匹配（MAE0）。会议M1与v2 p2完全匹配（MAE0）；v1 p2缺少MV/TP/RM行标签，且子图排版不同，故M不回溯v1。已实际查看两历史PDF、会议版图注与正文及导出。Hao Tan与Adobe作者身份一致；真实绘图日期未知。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Hao Tan；12273；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[49, 72, 294, 490]，PDF点、左上角原点。

Figure 1. Our Carve3D algorithm steadily improves the 3D con-
sistency of a multi-view diffusion model and the resulting quality
of the NeRF and the mesh, without sacrificing its image-prompt
alignment, texture details, or realism. Here, we show 3 testing-
set results (in 3 rows, numbered as 1-3, separated by dotted lines)
from the finetuning process (epoch 0, 28, and 55 in 3 columns).
Each row includes the generated multi-view images (denoted as
MV), the reconstructed NeRF and extracted mesh (denoted as RM)
and the text prompt (denoted as TP). The inconsistencies in the
multi-view images, e.g. the facing direction of the shopping cart,
the position of the octopus arms, and the position of the pencils,
lead to artifacts in the NeRF and the mesh (highlighted in red).

最早核实视觉日期：2024-12-05；arXiv v1：2024-12-05；CVPR 2025 fixed official PDF

已打开arXiv v1(2024-12-05)第3页。图2主图对象坐标及原始像素尺寸相同；所有模块、连线、文字和输出相同，差异图为压缩/栅格化误差(MAE=2.285)，不是新增图形内容。最早匹配v1。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Oisin Mac Aodha；18397；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[68, 74, 550, 242]，PDF点、左上角原点。

Figure 2. CrossSDF takes a set of planar cross-sections as input. These cross-sections result in contours which denote the surface
boundaries of the target geometry of interest (left, black lines) and each induces a 2D signed distance field (SDF) in its respective plane.
From this, we generate a set of planar sample points Ω and their 2D SDF labels. During training, points are encoded using our hybrid
pl
encoder before being passed to the SDF network M for prediction. At each iteration we create a set of 3D samples Ω to apply
SDF reg
volumetric regularization. The combination of our novel sampling, loss function, and hybrid encoding results in a high quality 3D SDF.

最早核实视觉日期：2024-12-05；arXiv v1：2024-12-05；CVPR 2025 fixed official PDF

已打开arXiv v1(2024-12-05)第1页并比较固定会议图1，完整图区域逐像素一致(MAE=0)。后续v2/v3不改变最早匹配时间。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Oisin Mac Aodha；18397；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[58, 218, 550, 402]，PDF点、左上角原点。

Figure 1. We propose CrossSDF, a novel approach for reconstructing a 3D signed-distance field from 2D cross-sections. The input is a
set of 2D cross-sections that sample an unobserved ground-truth geometric object by planar intersection (denoted as black lines overlayed
on the ground truth Alveolis structure on the left). CrossSDF (middle) accurately reconstructs thin structures without breakages, over-
smoothing, or cross-sectional artifacts observed in competing methods (right).

最早核实视觉日期：2024-03-30；arXiv v1：2024-03-30；CVPR 2024 fixed official PDF

已读正文/图注、查看固定源PDF，并核对作者单位。 已逐张查看完整源页、历史目标图与导出图；GenNBV首图历史页位置下移5点已对应；DiffHuman首图选择明确诊断失败的Fig3并保留全部列标签；图内元素完整、图外正文未混入。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Cristian Sminchisescu；24653；快照2026-10-22。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[51, 62, 542, 158]，PDF点、左上角原点。

Figure 2. Method overview. We use a diffusion probabilistic model [18] to predict a distribution over plausible 3D reconstructions
conditioned on a single RGB image. During training, we predict noise-dependent pixel-aligned features g (t) (x , I) given a noisy observation
Θ t
set x consisting of front/back albedo, depth and normal renders, and an RGB image I. These features condition an SDF f (t) , which is
t Θ
dependent on both x and I. f (t) and g (t) are neural networks that define an implicit surface S (t) (x , I). Then, we obtain an estimate of the
t Θ Θ Θ t
denoised observation set x(t) by rendering S (t) . We may additionally produce a shaded image C(t) by applying a pixel-wise noise-dependent
0Θ Θ
shading network s(t) . During inference, we can sample trajectories over observation sets x ∼ p (x |I) by computing and rendering
Θ 0:T Θ 0:T
S (t) (x , I) in each denoising step. Our final 3D samples S ∼ p (S |I) are obtained as the final reconstruction S = S (1) (x , I). To mitigate
Θ t Θ Θ 1
the computational cost of rendering an implicit surface in every step, we train a “generator” network h(t) that imitates rendering by directly
Θ
mapping g (t) (x , I) to x(t) . During inference, we denoise using h(t) and only explicitly compute the 3D reconstruction in the last step.
Θ t 0Θ Θ

最早核实视觉日期：2024-03-30；arXiv v1：2024-03-30；CVPR 2024 fixed official PDF

已读正文/图注、查看固定源PDF，并核对作者单位。 已逐张查看完整源页、历史目标图与导出图；GenNBV首图历史页位置下移5点已对应；DiffHuman首图选择明确诊断失败的Fig3并保留全部列标签；图内元素完整、图外正文未混入。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Cristian Sminchisescu；24653；快照2026-10-22。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[50, 56, 547, 276]，PDF点、左上角原点。

Figure 3. Qualitative comparison against deterministic monocular 3D human reconstruction methods [5, 10] that predict geometry,
surface albedo and shaded colour. PHORHUM [5] (retrained on our dataset) outputs good front predictions, but exhibits over-smooth, flat
geometry and blurry colours on the back. S3F [10] yields more detailed geometry, but colours are still often blurry. Moreover, both these
methods occasionally paste the front colour predictions onto the back incorrectly (see row 3). Our method outputs multiple diverse samples,
with a greater level of geometric detail and colour sharpness in uncertain regions, that are consistent with the input image after shading.

最早核实视觉日期：2024-02-25；arXiv v1：2024-02-25；CVPR 2024 fixed official PDF

已读正文/图注、查看固定源PDF，并核对作者单位。 已逐张查看完整源页、历史目标图与导出图；GenNBV首图历史页位置下移5点已对应；DiffHuman首图选择明确诊断失败的Fig3并保留全部列标签；图内元素完整、图外正文未混入。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Jiangmiao Pang；26222；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[50, 74, 542, 219]，PDF点、左上角原点。

Figure 2. Overview of our proposed framework GenNBV. Our end-to-end policy takes the historical multi-source observations as input,
transforms them into a more informative scene representation, and produces the next viewpoint position. A reward signal will be returned
at training time to optimize the end-to-end policy for maximizing the expected cumulative reward in one episode. Specifically, the signal
is the increased coverage ratio after collecting a new viewpoint.

最早核实视觉日期：2024-02-25；arXiv v1：2024-02-25；CVPR 2024 fixed official PDF

已读正文/图注、查看固定源PDF，并核对作者单位。 已逐张查看完整源页、历史目标图与导出图；GenNBV首图历史页位置下移5点已对应；DiffHuman首图选择明确诊断失败的Fig3并保留全部列标签；图内元素完整、图外正文未混入。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Jiangmiao Pang；26222；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[49, 201, 539, 320]，PDF点、左上角原点。

Figure 1. To determine the best view for 3D reconstruction, previous methods only chose from hand-crafted action space or based on
object-centric capturing, lacking the ability to generalize to unforeseen scenes (Left). With our end-to-end trained, generalized free-space
policy, it can generalize to unseen objects, enabling the captured drone to image from any viewpoint (Right).

最早核实视觉日期：2024-12-19；arXiv v1：2024-12-19；CVPR 2025 fixed official PDF

会议图区与arXiv v1按同坐标裁剪并比较；正文和图注参考文献编号变化不被复制到图内。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Jia-Bin Huang；38451；快照2026-10-21。

绘图工具：unknown。已读图注与相关正文；未找到该插图实际制作工具声明。

裁剪坐标：[58, 75, 294, 432]，PDF点、左上角原点。

Figure 2. Method overview. We first apply a relighting diffusion
modelthatconvertsN imagesI ,...,I withknowncameraposes
1 N
π , ..., π , captured under extremely different illuminations, to a
1 N
set of images with the same poses, but rendered under the illumi-
nation of the reference image I (highlighted in orange). We then
1
optimize a neural radiance field to obtain a consistent 3D repre-
sentation with a novel per-image shading embedding, which can
be used to render new views of the scene from unobserved poses.

最早核实视觉日期：2024-12-19；arXiv v1：2024-12-19；CVPR 2025 fixed official PDF

会议图区与arXiv v1按同坐标裁剪并比较；正文和图注参考文献编号变化不被复制到图内。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Jia-Bin Huang；38451；快照2026-10-21。

绘图工具：unknown。已读图注与相关正文；未找到该插图实际制作工具声明。

裁剪坐标：[58, 273, 554, 445]，PDF点、左上角原点。

Figure 1. 3D reconstruction under extreme illumination variation. We propose a method for 3D reconstruction from a set of images
captured under strongly varying illumination. Our method recovers high-fidelity appearance details including specular highlights that prior
state-of-the-artapproachescannotrecover(topbaseline: NeRF-Casting[47]withappearanceembeddings,bottombaseline: NeROIC[25]).

最早核实视觉日期：2024-06-11；arXiv v1：2024-06-11；CVPR 2024 fixed official PDF

已读正文/图注、查看固定源PDF，并核对作者单位。 已逐张查看完整源页、历史目标图与导出图；GenNBV首图历史页位置下移5点已对应；DiffHuman首图选择明确诊断失败的Fig3并保留全部列标签；图内元素完整、图外正文未混入。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Zhanyu Ma；11953；快照2026-10-22。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[49, 71, 547, 237]，PDF点、左上角原点。

Figure 4. Pipeline of our NeRSP.

最早核实视觉日期：2024-06-11；arXiv v1：2024-06-11；CVPR 2024 fixed official PDF

已读正文/图注、查看固定源PDF，并核对作者单位。 已逐张查看完整源页、历史目标图与导出图；GenNBV首图历史页位置下移5点已对应；DiffHuman首图选择明确诊断失败的Fig3并保留全部列标签；图内元素完整、图外正文未混入。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Zhanyu Ma；11953；快照2026-10-22。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[308, 72, 547, 366]，PDF点、左上角原点。

Figure 3. Ambiguity of determining 3D postions under sparse
views with geometric and photometric cues.

最早核实视觉日期：2025-03-11；arXiv v1：2025-03-11；CVPR 2025 fixed official PDF

已查看唯一v1（2025-03-11）与会议版整体框架，布局、配色和标签一致；轻微栅格差异。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Xiao Xiang Zhu；51863；快照2026-10-22。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[59, 71, 553, 278]，PDF点、左上角原点。

Figure 2. Architecture of PaCo. Starting from an incomplete point cloud, we hierarchically encode the points into plane proxies with a
lookup table from segmentation. These proxies inform the proxy generator in producing a set of proxy proposals. The generated proxies
are then passed to the parameter estimator and the point distributor for primitive estimation. Finally, the primitive selector identifies a
subset of primitives to form the parametric completion, with selection optimized via bipartite matching. The resulting primitives enable
direct assembly for polygonal surface reconstruction. Points are colored according to their planar primitives.

最早核实视觉日期：2025-03-11；arXiv v1：2025-03-11；CVPR 2025 fixed official PDF

已查看唯一v1（2025-03-11）与会议版目标图，元素、构图和标签一致；轻微栅格压缩差异。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Xiao Xiang Zhu；51863；快照2026-10-22。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[322, 203, 553, 419]，PDF点、左上角原点。

Figure 1. Unlike conventional point cloud completion that aims
to recover individual points (middle, non-planar points shown in
red), parametric completion recovers primitive parameters along
with corresponding inlier points. PaCo recovers planar structures
(right, each color denoting a planar primitive), directly enabling
high-quality polygonal surface reconstruction.

最早核实视觉日期：2024-12-04；arXiv v1：2024-12-04；CVPR 2025 fixed official PDF

已打开仅有的arXiv v1(2024-12-04)第4页。图3经统一比例缩放后，所有模块、箭头、渲染图与输入照片一致；该页23个共享嵌入图像解码像素相同，归一化裁图差异仅边缘栅格化，未发现场景替换。最早匹配v1。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Yujun Shen；15694；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[75, 86, 540, 302.5]，PDF点、左上角原点。

Figure 3. Illustration of our proposed PlanarSplatting. Given a set of posed multi-view images of indoor scenes, our method renders
depth and normal maps from 3D plane primitives. Then, with the supervision of monocular cues, these 3D plane primitives are gradually
optimized to recover the scene geometry and finally merged to get the planar reconstruction result.

最早核实视觉日期：2024-12-04；arXiv v1：2024-12-04；CVPR 2025 fixed official PDF

已打开仅有的arXiv v1(2024-12-04)第4页，历史图4对应会议图5。高斯/平面权重图、迭代标签、颜色图例均一致，只随版式缩放并改图号。保存完整历史图区域。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Yujun Shen；15694；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[325, 350, 543, 435.5]，PDF点、左上角原点。

Figure 5. Illustration of the proposed plane splatting function.
Naive Gaussian Splatting can not effectively approximate the
boundary of our rectangular plane primitive (shown in black
dashed border). In contrast, our proposed plane splatting function
can approximate the boundary of the rectangle as the number of
iterations increases, allowing our 3D planar primitives to better fit
the surface of the scene.

最早核实视觉日期：2025-03-21；arXiv v1：2025-03-21；CVPR 2025 fixed official PDF

已对照固定会议版本与arXiv v1，完整布局、标签、元素与连线匹配；v1为目前最早可核实的该视觉版本。已在PDF首页核对作者及单位。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Jerome Revaud；13410；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[58, 70, 300, 166]，PDF点、左上角原点。

Figure 2. Overview of the model architecture. Following
DUSt3R [75], images are encoded then decoded with a ViT back-
bone into pointmaps, from which focals, depthmaps and relative
pose can be extracted. Pow3R introduces optional inputs to guide
the regression with prior knowledge about the camera intrinsics
and depth (fed into the encoder) and the pose (into the decoder).

最早核实视觉日期：2025-03-21；arXiv v1：2025-03-21；CVPR 2025 fixed official PDF

已对照固定会议版本与arXiv v1，完整布局、标签、元素与连线匹配；v1为目前最早可核实的该视觉版本。已在PDF首页核对作者及单位。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Jerome Revaud；13410；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[58, 70, 555, 186]，PDF点、左上角原点。

Figure 4. Native resolution example. Methods like DUSt3R require centered principal point and would thus downscale the image to its
training resolution, leading to blurry prediction (top). With camera intrinsics input, we can process any crop in the image leading to native
resolution prediction (bottom row), for example by dividing the image into 4 crops, that are independently process as the first input image,
while the second input image of the network is set to the downscaled image.

最早核实视觉日期：2026-03-02；arXiv v1：2026-03-02；CVPR 2026 fixed official PDF

已查看固定会议PDF、图注和相关正文，核对唯一/最早v1目标图；源裁图与v1逐像素一致，MAE=0。保存v1及对应完整裁图。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Qi Wang；21434；快照2026-10-22。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[108, 72, 505, 303]，PDF点、左上角原点。

Figure 2. Overview of our framework at incremental step t. Given an image x, the query function generates a content-aware query
representation by adaptively computing query weights via a ranking function F and performing weighted aggregation. Subsequently,
ψ
prompts are retrieved from both the shared and private pool and injected into the decoder layer. In parallel, the teacher model Φ
t−1
generates a set of candidate bounding boxes, where potentially valuable ones are projected into the feature space to compute their similarity
with class prototypes. This process yields a set of refined, high-quality pseudo-labels to guide the training of the student model Φ .
t

最早核实视觉日期：2026-03-02；arXiv v1：2026-03-02；CVPR 2026 fixed official PDF

已查看固定会议PDF、图注和相关正文，核对唯一/最早v1目标图；源裁图与v1逐像素一致，MAE=0。保存v1及对应完整裁图。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Qi Wang；21434；快照2026-10-22。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[331, 232, 539, 545]，PDF点、左上角原点。

Figure 1. Comparison of prompt-based methods. (a) Use task
ID to isolate prompt, which forces task-general prompt to be re-
learned at each task. (b) Missing annotations cause prompt tokens
to drift. (c) The shared prompt pool continuously optimizes task-
general prompt, and category prototypes guide the generation of
pseudo labels for old categories.

最早核实视觉日期：2026-02-24；arXiv v1：2026-02-24；CVPR 2026 fixed official PDF

已核对固定会议PDF和arXiv v1的完整目标图；构图、文字、配色、元素和连线一致，v1为目前最早可核实的视觉版本。作者和机构已在PDF首页与公开引用主页交叉核对。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：C. V. Jawahar；32310；快照2026-10-21。

绘图工具：Gemini 2.5 Pro（仅图中说明性场景图像；不推断整图布局制作工具）。固定会议PDF第2页脚注1明确：“Images are generated using Gemini 2.5 Pro for illustration purposes.” 图1、图3图注带脚注1；只将披露归于说明性图像，不推断整张框架图由生成模型制作。证据：papers/2602.20985_conference.pdf，第2页；quota_review/text/2602.20985_conference.txt。

裁剪坐标：[69, 69, 541, 336.5]，PDF点、左上角原点。

Figure 3. EW-DETR framework for Evolving-World Object Detection. An input image is processed by a frozen backbone and
transformer encoder–decoder equipped with aggregate (cid:0) ∆Wt−1 (cid:1) and task-specific (cid:0) ∆Wt (cid:1) LoRA adapters, yielding class-agnostic
agg task
query features. These queries are reparameterised by the Query-Norm Objectness Adapter and passed through a classification head and
an objectness head, whose outputs are combined by Entropy-aware Unknown Mixing module to produce calibrated class scores, while a
localisation head predicts bounding boxes to form final detections across tasks1 .

最早核实视觉日期：2026-02-24；arXiv v1：2026-02-24；CVPR 2026 fixed official PDF

已核对固定会议PDF和arXiv v1的完整目标图；构图、文字、配色、元素和连线一致，v1为目前最早可核实的视觉版本。作者和机构已在PDF首页与公开引用主页交叉核对。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：C. V. Jawahar；32310；快照2026-10-21。

绘图工具：Gemini 2.5 Pro（仅图中说明性场景图像；不推断整图布局制作工具）。固定会议PDF第2页脚注1明确：“Images are generated using Gemini 2.5 Pro for illustration purposes.” 图1、图3图注带脚注1；只将披露归于说明性图像，不推断整张框架图由生成模型制作。证据：papers/2602.20985_conference.pdf，第2页；quota_review/text/2602.20985_conference.txt。

裁剪坐标：[95, 184, 514, 370]，PDF点、左上角原点。

Figure 1. Overview of Evolving World Object Detection (EWOD). Top (Training): Sequential tasks introduce disjoint classes K across
t
shifting domains D (Day → Night → Fog). Bottom (Deployment): The detector must (i) identify unseen objects U without supervision;
t t
(ii) retain knowledge of all prior classes across all domains; (iii) incrementally learn former unknowns when their labels are revealed later
(e.g., U → K ) and (iv) achieve all these objectives without revisiting any previous data. Best viewed in colour with zoom1 .
2 3

最早核实视觉日期：2026-02-27；arXiv v1：2026-02-27；CVPR 2026 fixed official PDF

已对照会议固定PDF与arXiv v1目标图，构图、文字和连线一致，导出像素一致；未发现更早公开版本。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Ying Fu；11391；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[317, 242, 553, 384]，PDF点、左上角原点。

Figure 1. The position of FAAFusion and FAA Head.

最早核实视觉日期：2026-02-27；arXiv v1：2026-02-27；CVPR 2026 fixed official PDF

已对照会议固定PDF与arXiv v1目标图，构图、文字和连线一致，导出像素一致；未发现更早公开版本。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Ying Fu；11391；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[58, 70, 295, 160]，PDF点、左上角原点。

Figure 2. Directional incoherence. High-level feature (c) has
strong semantics and large receptive fields but captures the coarse
orientation. Forexample,from(c)wecanonlyinferthattheobject
lieshorizontally. Yetmanydown-samplingsblurspatialdetails, so
the orientation signal is low-frequency and vague. Low-level fea-
ture (b) keeps rich edges, corners, and textures, giving clear and
high-frequency orientation cues.

最早核实视觉日期：2026-03-25；arXiv v1：2026-03-25；CVPR 2026 fixed official PDF

v1与会议版框架的全部模块、图例、字体及连线对应；会议版整体缩小并居中，未发现视觉重绘。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Dacheng Tao；205014；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[106, 72, 504, 268]，PDF点、左上角原点。

Figure 2. HeROD pipeline. Text and image features are encoded, and spatial/semantic reasoning priors are derived from the phrase and
image. These priors are injected into reference generation, final prediction, and training loss, guiding the detector toward plausible regions
for more data-efficient learning.

最早核实视觉日期：2026-03-25；arXiv v1：2026-03-25；CVPR 2026 fixed official PDF

v1与会议图的完整裁剪像素一致；采用v1公开日期2026-03-25。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Dacheng Tao；205014；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[317, 245, 551, 420]，PDF点、左上角原点。

Figure 1. Conceptual illustration. (a) Without priors, grounding
detectors trained with limited labels often ignore spatial/semantic
cues (e.g., “on the left”). (b) HeROD injects heuristic-inspired
reasoning priors (spatial + semantic), guiding the detector toward
correct localization and improving data-efficient learning.

最早核实视觉日期：2026-02-06；arXiv v1：2026-02-06；arXiv v1 fixed

已读固定v1图注及引用正文，查看完整页面与300dpi导出；首版包含同一图，v1早于S4。LAB-Det作者Dacheng Tao同名及邮箱由其共同作者公开论文核对（https://gcatnjust.github.io/ChenGong/paper/luo_tpami24.pdf）。SeSAM作者机构页面登记2026预印本，TMLR公开周报出现于2026-05-10，未发现早于v1版本；不将后续审稿时间代替图公开日期。实际创作日期unknown。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Wen Li；14507；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[66, 67, 531, 214]，PDF点、左上角原点。

Figure 2. An overview of the proposed RSCN. For every batch, source-domain images with labeled foreground objects and target-domain
images without foreground objects are fed to the detector. The background prototypes are aligned with the BPA objective in Eq. 5. RSH in
Eq. 6 keeps the relative geometric relationship for the source and target background prototypes to the shared foreground prototype anchors.
A frozen reference detector is utilized to maintain the source-domain structure in order to avoid the feature collapse with SSP in Eq. 7.

最早核实视觉日期：2026-02-06；arXiv v1：2026-02-06；arXiv v1 fixed

已读固定v1图注及引用正文，查看完整页面与300dpi导出；首版包含同一图，v1早于S4。LAB-Det作者Dacheng Tao同名及邮箱由其共同作者公开论文核对（https://gcatnjust.github.io/ChenGong/paper/luo_tpami24.pdf）。SeSAM作者机构页面登记2026预印本，TMLR公开周报出现于2026-05-10，未发现早于v1版本；不将后续审稿时间代替图公开日期。实际创作日期unknown。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Wen Li；14507；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[308, 174, 541, 397]，PDF点、左上角原点。

Figure 1. Illustration of Instance-Free DAOD. The orange time-
line indicates that target-domain instances appear only after a
long waiting period and require costly manual screening. Tradi-
tional DAOD (top) relies on target images containing foreground
instances, whereas Instance-Free DAOD (bottom) enables transfer-
ring using only background-only target images.

最早核实视觉日期：2026-03-03；arXiv v1：2026-03-03；ECCV 2026 fixed official PDF

已查看arXiv v1和会议版的两张目标图；F由第6页移至第5页，构图/文字/连线未见变化。历史提交记录另含v2/v3，当前视觉已在v1存在。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Alessio R Lomuscio；11623；快照2026-10-21。

绘图工具：unknown。已读图注与相关正文；未找到该插图实际制作工具声明。

裁剪坐标：[134, 116, 481, 210]，PDF点、左上角原点。

Fig.2: Pipeline of our object detection (OD) verification framework, combining verifier
bounds, candidate selection, and optimal IoU bound derivation to reach ROBUST,
NONROBUST or UNKNOWN conclusions.

最早核实视觉日期：2026-03-03；arXiv v1：2026-03-03；ECCV 2026 fixed official PDF

已查看arXiv v1和会议版的两张目标图；F由第6页移至第5页，构图/文字/连线未见变化。历史提交记录另含v2/v3，当前视觉已在v1存在。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Alessio R Lomuscio；11623；快照2026-10-21。

绘图工具：unknown。已读图注与相关正文；未找到该插图实际制作工具声明。

裁剪坐标：[229, 114, 387, 187]，PDF点、左上角原点。

Fig.1: A counterexample from robustness verification of a YOLOv3 model on a runway
detection task. Left: original image; right: perturbed image. Green boxes: ground truth,
blue box: original model prediction (IoU = 0.89), red box: prediction on the perturbed
image (IoU = 0.11).

最早核实视觉日期：2026-02-06；arXiv v1：2026-02-06；arXiv v1 fixed

已读固定v1图注及引用正文，查看完整页面与300dpi导出；首版包含同一图，v1早于S4。LAB-Det作者Dacheng Tao同名及邮箱由其共同作者公开论文核对（https://gcatnjust.github.io/ChenGong/paper/luo_tpami24.pdf）。SeSAM作者机构页面登记2026预印本，TMLR公开周报出现于2026-05-10，未发现早于v1版本；不将后续审稿时间代替图公开日期。实际创作日期unknown。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Dacheng Tao；205014；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[62, 74, 552, 337]，PDF点、左上角原点。

Figure 2. Overview of LAB-Det. Top: exemplar-to-language projection. A single annotated exemplar is segmented by SAM and described
by DAM under a domain-aware prompt, yielding natural-language phrases (e.g., “rough texture”). Bottom: these phrases condition a
frozen detector (e.g., Grounding DINO) to generate candidate boxes and phrase scores. Category scores are obtained by averaging phrase
scores, and an optional BLIP-based calibration refines small or ambiguous detections. The entire pipeline is training-free and interpretable.

最早核实视觉日期：2026-02-06；arXiv v1：2026-02-06；arXiv v1 fixed

已读固定v1图注及引用正文，查看完整页面与300dpi导出；首版包含同一图，v1早于S4。LAB-Det作者Dacheng Tao同名及邮箱由其共同作者公开论文核对（https://gcatnjust.github.io/ChenGong/paper/luo_tpami24.pdf）。SeSAM作者机构页面登记2026预印本，TMLR公开周报出现于2026-05-10，未发现早于v1版本；不将后续审稿时间代替图公开日期。实际创作日期unknown。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Dacheng Tao；205014；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[317, 248, 553, 471]，PDF点、左上角原点。

Figure 1. Challenge and conceptual comparison of training-free 1-
shot domain generalization for object detection. (a) In specialized
domains (e.g., underwater or industrial), data can be very scarce
(e.g., only one labeled exemplar per class) and object boundaries
are ambiguous, making general-domain detectors fail under severe
domain gaps. (b) Training-based approaches rely on fine-tuning
with gradient updates, which incur extra cost, overfit under 1-shot
supervision, and remain unstable in high-IB domains. In contrast,
LAB-Det leverages language as a domain-invariant bridge: sup-
port exemplars are re-expressed as descriptive text and injected
into frozen vision–language models, enabling training-free, inter-
pretable, and robust target-domain detection.

最早核实视觉日期：2026-03-22；arXiv v1：2026-03-22；CVPR 2026 fixed official PDF

已打开会议PDF，阅读图注及引言/方法，查看完整源页及300dpi输出。唯一v1公开于2026-03-22；源图与v1完整裁图逐像素相同(MAE=0)。作者Tianjin University身份匹配Wei Feng公开Scholar11058次。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Wei Feng；11058；快照2026-10-22。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[82, 72, 525, 252]，PDF点、左上角原点。

Figure 2. Illustration of the training process of NoOVD. We use K-FPN to extract the pyramid embeddings from frozen CLIP to identify
latent novel-category objects. Besides the image-text alignment with L , we also align the features of the RoI head with the features
cons
from K-FPN via knowledge self-distillation with L .
kd

最早核实视觉日期：2026-03-22；arXiv v1：2026-03-22；CVPR 2026 fixed official PDF

已打开会议PDF，阅读图注及引言/方法，查看完整源页及300dpi输出。唯一v1公开于2026-03-22；源图与v1完整裁图逐像素相同(MAE=0)。作者Tianjin University身份匹配Wei Feng公开Scholar11058次。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Wei Feng；11058；快照2026-10-22。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[329, 276, 541, 450]，PDF点、左上角原点。

Figure 1. Training process for OVD using frozen CLIP. (a) Com-
monly used training process, (b) Our training process.

最早核实视觉日期：2026-03-10；arXiv v1：2026-03-10；arXiv v1 fixed

已查看v1、v2及会议版：会议图3相对v1等比放大，会议图2相对v1/v2更改了连接节点等细节；本条采集的是原始v1视觉，不给会议重绘版使用旧日期。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Xiaoyan Sun；15442；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[134, 116, 479, 294]，PDF点、左上角原点。

Fig. 2: The main architecture of our proposed RiO-DETR. The framework highlights
three key components: (A) Content-Driven Angle Estimation, (B) Rotation-Rectified
Orthogonal Attention, and (C) Decoupled Periodic Refinement.

最早核实视觉日期：2026-03-10；arXiv v1：2026-03-10；arXiv v1 fixed

已查看v1、v2及会议版：会议图3相对v1等比放大，会议图2相对v1/v2更改了连接节点等细节；本条采集的是原始v1视觉，不给会议重绘版使用旧日期。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Xiaoyan Sun；15442；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[161, 115, 453, 215]，PDF点、左上角原点。

Fig. 3: An intuitive illustration of (D) Decoupled Periodic Refinement and (E) Oriented
Dense O2O.

最早核实视觉日期：2026-03-17；arXiv v1：2025-10-01；ECCV 2026 fixed official PDF

已查看v1、v2及会议版；v1图1检测标签较小，v1框架的提示选择区域/标注与会议版不同。v2已包含会议版的标签与框架设计，会议版F仅排版缩放；以v2的2026-03-17日期登记当前视觉，不套用v1日期。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Eric Granger；14279；快照2026-10-21。

绘图工具：unknown。已读图注与相关正文；未找到该插图实际制作工具声明。

裁剪坐标：[137, 114, 476, 237]，PDF点、左上角原点。

Fig.3: Overview of VLOD-TTA. Given an input image and a pool of class
prompts, the text encoder produces embeddings that interact with region proposals
through the vision–language head to compute similarity scores. IPS selects the top-ρ
prompts and averages their scores to obtain per-proposal class scores. IWE then com-
bines per-proposal entropy with IoU-based weights to form the adaptation objective.

最早核实视觉日期：2026-03-17；arXiv v1：2025-10-01；ECCV 2026 fixed official PDF

已查看v1、v2及会议版；v1图1检测标签较小，v1框架的提示选择区域/标注与会议版不同。v2已包含会议版的标签与框架设计，会议版F仅排版缩放；以v2的2026-03-17日期登记当前视觉，不套用v1日期。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Eric Granger；14279；快照2026-10-21。

绘图工具：unknown。已读图注与相关正文；未找到该插图实际制作工具声明。

裁剪坐标：[142, 114, 474, 208]，PDF点、左上角原点。

Fig.1: Motivation for IWE. Left→right: (i) Raw predictions before thresholding
for two classes — a true positive Person (red) (bbox cluster size = 167, max score
= 0.14) and a false positive Dog (blue) (bbox cluster size = 45, max score = 0.15);
(ii) In ZS, all scores remain below the detection threshold, causing a missed detection;
(iii) Standard entropy minimization over-confidently sharpens scores, resulting in a dog
false positive; and (iv) IWE focuses updates on dense bbox clusters and suppresses
isolated boxes, producing only the correct person detection.

最早核实视觉日期：2026-04-09；arXiv v1：2026-04-09；CVPR 2026 fixed official PDF

已逐张查看arXiv v1和会议PDF，两张视觉一致；保存原始PDF与匹配裁图。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Yanyun Qu；11764；快照2026-10-21。

绘图工具：unknown。已读图注与相关正文；未找到该插图实际制作工具声明。

裁剪坐标：[58, 71, 554, 214]，PDF点、左上角原点。

Figure 2. Overview of the proposed method. The pipeline begins by computing the cosine similarity (Cos.) between vision-language
features to obtain logits, followed by non-maximum suppression (NMS) and normalization (Norm.) to filter low-confidence patches. The
core step involves calculating the distribution discrepancy between the normalized logits and the degenerate distribution map. This solved
discrepancy is then upsampled to the original image resolution via joint bilateral upsampling (JBU) to obtain the discrepancy map. Finally,
the semantic map is generated by applying an argmax operation to the discrepancy map.

最早核实视觉日期：2026-04-09；arXiv v1：2026-04-09；CVPR 2026 fixed official PDF

已逐张查看arXiv v1和会议PDF，两张视觉一致；保存原始PDF与匹配裁图。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Yanyun Qu；11764；快照2026-10-21。

绘图工具：unknown。已读图注与相关正文；未找到该插图实际制作工具声明。

裁剪坐标：[316, 241, 555, 391]，PDF点、左上角原点。

Figure 1. Compared with the logits-optimization methods seeking
optimal solutions, we reformulate the problem as an analytical so-
lution form. This reformulation confers three key advantages: in-
dependence from GT annotations, elimination of time-consuming
training, and freedom from model-specific modulation.

最早核实视觉日期：2026-04-13；arXiv v1：2026-04-13；arXiv v1 fixed

已读固定v1图注及引用正文，查看完整页面与300dpi导出；首版包含同一图，v1早于S4。LAB-Det作者Dacheng Tao同名及邮箱由其共同作者公开论文核对（https://gcatnjust.github.io/ChenGong/paper/luo_tpami24.pdf）。SeSAM作者机构页面登记2026预印本，TMLR公开周报出现于2026-05-10，未发现早于v1版本；不将后续审稿时间代替图公开日期。实际创作日期unknown。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Bernt Schiele；131438；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[73, 82, 540, 200]，PDF点、左上角原点。

Figure 3: SeSAM framework for coarse sup ervision. First, pseudo-labels for training images are
generated by the teacher model. Next, these pseudo-labels are then refined following the key steps 1-3,
namely instance separation, point sampling and mask selection (see Section 3). To improve mask quality,
we augment coarse labels with high-confidence pseudo-labels before applying SAM (step 0). The resulting
refined labels, called SeSAM labels, are then used to train the student network.

最早核实视觉日期：2026-04-13；arXiv v1：2026-04-13；arXiv v1 fixed

已读固定v1图注及引用正文，查看完整页面与300dpi导出；首版包含同一图，v1早于S4。LAB-Det作者Dacheng Tao同名及邮箱由其共同作者公开论文核对（https://gcatnjust.github.io/ChenGong/paper/luo_tpami24.pdf）。SeSAM作者机构页面登记2026预印本，TMLR公开周报出现于2026-05-10，未发现早于v1版本；不将后续审稿时间代替图公开日期。实际创作日期unknown。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Bernt Schiele；131438；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[70, 91, 540, 249]，PDF点、左上角原点。

Figure 2: SAM challenges. Left: As an instance-based model, SAM performs poorly when segmenting a
class comprising multiple instances, see the class “person”. Splitting the class mask into individual instances
addresses this challenge. Middle: Confidence-based sampling (Kweon & Yoon, 2024) yields a suboptimal
mask; in contrast, our proposed sampling strategy achieves comprehensive mask coverage, resulting in im-
proved segmentation. Right: SAM score-based mask selection may not give the best desired segmentation
(see mask in red box). Our mask selection strategy using intersection with the coarse mask determines the
desired mask granularity (see mask in green box).

最早核实视觉日期：2026-03-03；arXiv v1：2026-03-03；arXiv v1 fixed source

固定arXiv v1作为本次采集原件；v1图形直接取自该版本。保留会议版差异与既知S3投稿记录，不把会议日期隐藏或修改。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Nicu Sebe；59898；快照2026-10-21。

绘图工具：unknown。已读图注、引言和方法；未获得插图制作工具声明。

裁剪坐标：[72, 71, 536, 341]，PDF点、左上角原点。

Figure 4. Overview of the proposed GKD framework. GKD comprises two major parts: domain-general distillation and task learning. In
the domain-general distillation stage, the student sequentially performs task-agnostic and domain-agnostic distillation, both via the Query-
based Soft Distillation mechanism. In the task learning stage, only the decoder is trained on source annotations, while the student encoder
is frozen to preserve the domain-general representations.

最早核实视觉日期：2026-03-03；arXiv v1：2026-03-03；arXiv v1 fixed source

固定arXiv v1作为本次采集原件；v1图形直接取自该版本。保留会议版差异与既知S3投稿记录，不把会议日期隐藏或修改。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Nicu Sebe；59898；快照2026-10-21。

绘图工具：unknown。已读图注、引言和方法；未获得插图制作工具声明。

裁剪坐标：[316, 260, 555, 335]，PDF点、左上角原点。

Figure 1. Comparison of Knowledge Distillation (KD) and our
proposed generalizable KD (GKD). Conventional KD preserves
accuracy within the same domain but overlooks generalization to
unseen domains.

最早核实视觉日期：2026-03-17；arXiv v1：2025-10-17；CVPR 2026 fixed official PDF

v1/v2框架使用无衬线字体，v3改为与会议版相同的衬线字体并调整细节；采用v3公开日期2026-03-17。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Xuelong Li；102448；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[58, 70, 554, 205.5]，PDF点、左上角原点。

Figure 3. Overall framework of the proposed Method. The Geometric Prior Enhancement module strengthens structural representa-
tions via visual–geometric fusion and transformer-based query refinement. The Semantic Alignment Injection mechanism align category
semantics with degraded underwater conditions.

最早核实视觉日期：2025-10-17；arXiv v1：2025-10-17；CVPR 2026 fixed official PDF

已比对v1、v2、v3；M图元素、字体、颜色与布局在v1已出现，仅在页面中纵向移动23.942pt；保存v1对应裁图，F则另用v3。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Xuelong Li；102448；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[326, 240, 541, 504]，PDF点、左上角原点。

Figure 1. The challenges of transferring OV instance segmen-
tation to underwater scenarios in terms of (a) datasets and (b-c)
methods, which have motivated the contributions of this study.

最早核实视觉日期：2026-03-28；arXiv v1：2026-03-28；arXiv v1 fixed

已打开完整v1源PDF并读图注、引言和方法引用段；逐张检查导出图。SegEarth-OV3固定2025-12-09 v1，后续2026-04-22 v2不用于本样本；作者Deyu Meng的西安交大身份与公开引用主页一致。Syn4Seg作者Ming-Ming Cheng的南开身份与公开引用主页一致。实际作图日期unknown。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Ming-Ming Cheng；85763；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[48, 56, 565, 313]，PDF点、左上角原点。

Fig. 2. (a), (b) Given several novel classes and their support set, we first use HDIG to synthesize high-quality, diverse novel class images proportional to the
base class images. Guided by support images, APE refines these synthesized images to generate accurate masks through two stages: Adaptive Pseudo-label
Filtering (APF) and Adaptive Pseudo-label Relabeling (APR). Specifically, APF filters noisy regions in the initial mask M to obtain M1 ; APR relabels these
unlabeled regions to yield M2 . Finally, SAM-based Boundary Refinement (SABR) further refines the mask boundaries, yielding the final M3 . (c), (d) illustrate
the details of APF and APR.

最早核实视觉日期：2026-03-28；arXiv v1：2026-03-28；arXiv v1 fixed

已打开完整v1源PDF并读图注、引言和方法引用段；逐张检查导出图。SegEarth-OV3固定2025-12-09 v1，后续2026-04-22 v2不用于本样本；作者Deyu Meng的西安交大身份与公开引用主页一致。Syn4Seg作者Ming-Ming Cheng的南开身份与公开引用主页一致。实际作图日期unknown。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Ming-Ming Cheng；85763；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[321, 176, 554, 377]，PDF点、左上角原点。

Fig. 1. Pipelines of classical GFSS works and our Syn4Seg framework.
(a) Classical GFSS methods follow three stages: base class learning, novel
class registration, and evaluation, where novel classes rely on limited manual
annotations, resulting in low diversity and constrained performance. (b) Our
Syn4Seg framework synthesizes diverse, realistic novel class images and
pseudo labels guided by support annotations. These synthesized samples,
together with base data, are used to train a semantic segmentation model,
enabling robust recognition of both base and novel classes during evaluation.

最早核实视觉日期：2026-03-23；arXiv v1：2026-03-23；CVPR 2026 fixed official PDF

已打开arXiv v1(2026-03-23)及固定CVPR 2026图2。完整裁图逐像素匹配(MAE=0)，主流程及PA/TLP子图均一致。已核查CVPR 2026投稿窗口同为S3。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Byeungwoo Jeon；11367；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[57, 70, 555, 208]，PDF点、左上角原点。

Figure 2. Our PEARL framework with align-then-propagate. Given an image I and a label set Y , a frozen ViT vision encoder yields
patch tokens. Procrustes alignment is inserted in the last self-attention block: for each head (J heads), queries and keys are weighted-
centered and the keys are orthogonally rotated toward the query subspace, producing geometry-corrected patch features. A frozen text
encoder maps Y to unit-norm prototypes T . Patch-text cosine similarity gives initial logits Z(cid:101) . Text-aware Laplacian propagation then
refines Z(cid:101) on a compact grid using a class graph G induced by T and confidence weights, and the refined scores are upsampled to full
resolution F . The final segmentation S is obtained by selecting, at each pixel, the class with the highest score.

最早核实视觉日期：2026-03-23；arXiv v1：2026-03-23；CVPR 2026 fixed official PDF

已打开arXiv v1(2026-03-23)及固定CVPR 2026图1。完整裁图逐像素匹配(MAE=0)，注意力图及所有标签一致。已核查CVPR 2026投稿窗口2025-10-16至2025-11-14 UTC，均为S3。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Byeungwoo Jeon；11367；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[316, 258, 555, 400]，PDF点、左上角原点。

Figure 1. Attention visualization at selected query points. We
show an input resized to 512 × 512 using the CLIP ViT-B/16 vi-
sion encoder. For each colored query, we compare the attention
maps produced by CLIP [51], NACLIP [26], and our PEARL.
CLIP exhibits diffuse and background-biased responses, while
NACLIP improves localization but often fragments objects into
isolated peaks. By performing Procrustes alignment in the last
self-attention block, PEARL yields compact and object-consistent
attention, suppresses background spill, and better preserves thin
parts, resulting in stable focus with reduced grid artifacts.

最早核实视觉日期：2026-01-23；arXiv v1：2026-01-23；arXiv v1 fixed

固定采用2026-01-23 arXiv v1；会议图1把Roughly Collinearity改为Rough Collinearity并微调子图比例，故不把会议视觉套用v1时间。v1两图已逐张对照，完整矢量裁剪与300dpi PNG已目检。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Xi Li；19440；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[59, 74, 554, 263]，PDF点、左上角原点。

Figure 2. The overview of REL-SF4PASS. Firstly, REL representation effectively represents the depth information by using ReD, EGVIA,
and LOA, which contains the 3D location and surface normal direction. Secondly, Our SMMF to get image regions from cylinder side
surface and uses a Gate Network to independently determine the fusion used in each region.

最早核实视觉日期：2026-01-23；arXiv v1：2026-01-23；arXiv v1 fixed

固定采用2026-01-23 arXiv v1；会议图1把Roughly Collinearity改为Rough Collinearity并微调子图比例，故不把会议视觉套用v1时间。v1两图已逐张对照，完整矢量裁剪与300dpi PNG已目检。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Xi Li；19440；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[318, 231, 554, 504]，PDF点、左上角原点。

Figure 1. Overview of ERP projection, cylindrical coordinate and
comparison of HHA and REL. Tˆ is the tangent line to circle O
θ,p p
at point p. Circle O lies in the plane perpendicular to gravity,
p
passes through point p, and its center lies on the cylindrical axis
L of ρθz . In (c), we compared how different structure informa-
cyl
tion are represented in REL and HHA.

最早核实视觉日期：2026-03-06；arXiv v1：2026-03-06；CVPR 2026 main-conference fixed open-access version

arXiv历史仅列v1；第3页图3与CVPR固定版相同区域渲染逐像素一致(MAE=0)。原PDF该完整框架本身为位图，裁剪PDF保留原位图，未声称将其矢量化。 日期字段采用arXiv提交历史中的版本日期；精确公开公告时刻未单独核查，距分界均有余量。仅指已核查来源中最早证据，不证明全网首次。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Bernt Schiele；131438；快照2026-10-21。

绘图工具：unknown。已读图注和相关正文；未获得明确的该论文插图制作工具声明。方法中使用生成模型不等于制图工具披露。

裁剪坐标：[315, 72, 555, 239]，PDF点、左上角原点。

Figure 3. Overview of the training pipeline. Our framework operates in two stages. Base Training (blue and green) establishes independent student-teacher setups for each modality using sparse supervision. Cross-Modal Consistency (orange) introduces our core contribution: bidirectional knowledge transfer where the teacher of one modality supervises the student of the other, weighted by our dual confidence mechanism leveraging prediction certainty and reconstruction quality.

最早核实视觉日期：2026-03-06；arXiv v1：2026-03-06；CVPR 2026 main-conference fixed open-access version

arXiv历史仅列v1；第2页图2与CVPR固定版相同区域渲染逐像素一致(MAE=0)。检查了目录、摘要历史及完整v1/会议PDF。 日期字段采用arXiv提交历史中的版本日期；精确公开公告时刻未单独核查，距分界均有余量。仅指已核查来源中最早证据，不证明全网首次。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Bernt Schiele；131438；快照2026-10-21。

绘图工具：unknown。已读图注和相关正文；未获得明确的该论文插图制作工具声明。方法中使用生成模型不等于制图工具披露。

裁剪坐标：[315, 72, 555, 195]，PDF点、左上角原点。

Figure 2. Conceptual overview of weakly-supervised segmentation approaches. (a) Traditional methods rely solely on sparse 2D annotations, limiting supervision propagation. (b) Our proposed method Rewis3d introduces a 3D branch, enforcing cross-modal consistency (CMC) between 2D predictions and 3D predictions from reconstructed geometry.

最早核实视觉日期：2025-12-02；arXiv v1：2025-12-02；CVPR 2026 fixed official PDF

已对照会议固定PDF与arXiv v1目标图，构图、文字和连线一致，导出像素一致；未发现更早公开版本。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Haohuan Fu；14800；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[58, 70, 555, 241]，PDF点、左上角原点。

Figure 2. The training process of SPG: each style reference guides the source image through a dedicated generator G to produce a style-
i
aware prompt. Prompts are merged and used to optimize a frozen privacy model. Each style-prompt generator consists of a modulation
network and a learnable prompt template, producing the final style-prompt via element-wise modulation.

最早核实视觉日期：2025-12-02；arXiv v1：2025-12-02；CVPR 2026 fixed official PDF

已对照会议固定PDF与arXiv v1目标图，构图、文字和连线一致，导出像素一致；未发现更早公开版本。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Haohuan Fu；14800；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[58, 70, 552, 231]，PDF点、左上角原点。

Figure 1. Overview of domain generalization settings under privacy constraints. (a) Standard DGSS requires backbone tuning. (b) Style-
fixed prompting is ineffective on unseen domains. (c) SAGE inference overcomes privacy limitations.

最早核实视觉日期：2025-12-09；arXiv v1：2025-12-09；arXiv v1 fixed

已打开完整v1源PDF并读图注、引言和方法引用段；逐张检查导出图。SegEarth-OV3固定2025-12-09 v1，后续2026-04-22 v2不用于本样本；作者Deyu Meng的西安交大身份与公开引用主页一致。Syn4Seg作者Ming-Ming Cheng的南开身份与公开引用主页一致。实际作图日期unknown。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Deyu Meng；49595；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[59, 73, 558, 307]，PDF点、左上角原点。

Figure 2. The overall inference pipeline of SegEarth-OV3. Given an input image and a list of text prompts, we leverage SAM 3’s decoupled
outputs. The pipeline involves: (1) instance aggregation to consolidate sparse object queries; (2) dual-head mask fusion to combine the
fine-grained instance details with the global coverage of the semantic head; and (3) presence-guided filtering (using the presence score) to
suppress false positives from absent categories. denotes the element-wise maximum operation, and denotes multiplication.

最早核实视觉日期：2025-12-09；arXiv v1：2025-12-09；arXiv v1 fixed

已打开完整v1源PDF并读图注、引言和方法引用段；逐张检查导出图。SegEarth-OV3固定2025-12-09 v1，后续2026-04-22 v2不用于本样本；作者Deyu Meng的西安交大身份与公开引用主页一致。Syn4Seg作者Ming-Ming Cheng的南开身份与公开引用主页一致。实际作图日期unknown。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Deyu Meng；49595；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[322, 71, 548, 206]，PDF点、左上角原点。

Figure 1. The Transformer decoder of SAM 3 excels at delineat-
ing countable objects but produces fragmented masks for amor-
phous regions, while the semantic head preserves continuity for
amorphous regions but lacks boundary precision for small targets.
(Left: Remote sensing image. Middle: Prediction of Transformer
Decoder. Right: Prediction of semantic segmentation head.)

最早核实视觉日期：2025-12-04；arXiv v1：2025-12-03；CVPR 2026 fixed official PDF

已查看v1、v2、v3和会议PDF：M与v1相同，仅整体下移13.948 PDF点；F在v1的公式/标签不同，v2已匹配会议视觉，使用v2日期。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Bohyung Han；31556；快照2026-10-21。

绘图工具：unknown。已读图注与相关正文；未找到该插图实际制作工具声明。

裁剪坐标：[77, 73, 538, 310]，PDF点、左上角原点。

Figure 2. Overview of our framework. (a) The supervision enhancement framework produces enhanced ground truth images by fusing frequency components from the original ground truth I_0^GT and its super-resolved variants I_i^GT using adaptive frequency masks M_i. (b) The conditional frequency mask generator constructs M_i by combining predefined masks R_b weighted with predicted coefficients c_{i,b}, followed by a softmax function. (c) The output refinement network (ORNet) is trained with this enhanced supervision to improve perceptual quality.

最早核实视觉日期：2025-12-03；arXiv v1：2025-12-03；CVPR 2026 fixed official PDF

已查看v1、v2、v3和会议PDF：M与v1相同，仅整体下移13.948 PDF点；F在v1的公式/标签不同，v2已匹配会议视觉，使用v2日期。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Bohyung Han；31556；快照2026-10-21。

绘图工具：unknown。已读图注与相关正文；未找到该插图实际制作工具声明。

裁剪坐标：[330, 227, 545, 454]，PDF点、左上角原点。

Figure 1. Visualization of our enhanced ground truth (GT). Our
enhanced GT not only demonstrates sharper text and superior per-
ceptualqualitybutalsomaintainssemanticconsistencywithrespect
to the original GT. Zoom in for better visualization.

最早核实视觉日期：2026-03-09；arXiv v1：2026-03-09；CVPR 2026 fixed official PDF

已对照会议固定PDF与arXiv v1目标图，构图、文字和连线一致，导出像素一致；未发现更早公开版本。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Ying Fu；11391；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[58, 70, 554, 296]，PDF点、左上角原点。

Figure 3. The overall architecture. We adopt a multi-scale encoder-decoder based on unmixing with several key modules: (a) Coarse-
to-Fine Deformable Aggregation module, (b) Spatial-Channel Abundance Cross-Attention module, and (c) Spatial-Channel Modulation
Fusion module to super-resolve high-quality HR HSI.

最早核实视觉日期：2026-03-09；arXiv v1：2026-03-09；CVPR 2026 fixed official PDF

已对照会议固定PDF与arXiv v1目标图，构图、文字和连线一致，导出像素一致；未发现更早公开版本。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Ying Fu；11391；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[58, 70, 294, 278]，PDF点、左上角原点。

Figure 2. The motivation. (c) The explicit alignment within the
SSCH [53] is prone to introducing spatial distortions and textural
artifacts into the warped image. (d) The result validates the princi-
plethatleveragingawell-aligned, high-fidelityabundanceenables
the reconstruction of the HR HSI from its LR counterpart.

最早核实视觉日期：2026-04-06；arXiv v1：2025-12-27；CVPR 2026 fixed official PDF

核查v1与v2：v1奖励输出为5.64/7.89/3.25，图例火/雪与文字关系、模块边框及CCL标题位置不同；会议图匹配2026-04-06 v2，不套用v1日期。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Zheng-Jun Zha；30479；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[63, 71, 555, 350]，PDF点、左上角原点。

Figure2. TheoverallpipelineoftheproposedFinPercep-RMandCo-evolutionaryCurriculumLearning(CCL)framework. FinPercep-RM
producesaFine-grainedPerceptualDegradationMapthatcapturesspatiallylocalizeddefectlikelihoodandintensity, andtherewardmodel
is progressively expanded from small-variance, coarse-grained rewards to large-variance, fine-grained signals. During training, under the
CCL mechanism, the Generator first learns with the coarse global reward from RM for a stable and easy initialization, and is then co-
0
evolutionarily guided by increasingly strict RM versions to enhance local fidelity and suppress reward hacking.
k

最早核实视觉日期：2025-12-27；arXiv v1：2025-12-27；CVPR 2026 fixed official PDF

v1与v2/会议版图1的视觉内容相同；v1整图页面纵坐标下移9.962pt，按对应坐标核对全部子图与文字；排版位置变化不改变图形时间。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Zheng-Jun Zha；30479；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[317, 260, 554, 416]，PDF点、左上角原点。

Figure 1. Motivation for FinPercep-RM and CCL. (a) Standard
IQAs lack fine-grained perception and struggle to penalize lo-
cal distortions, while Ours aligns with human judgment (User
Study). (b) The training curves illustrate the stability-robustness
dilemma: baseline IQA rewards (blue/purple) converge quickly,
while FinPercep-RM (light blue) is oscillatory and unstable. Our
completemethod(FinPercep-RMw/CCL,orange)achievesstable
and optimal convergence. (c) Visualization of Reward Hacking:
baseline rewards (W/ CLIP-IQA, W/ MAN-IQA) produce local
artifacts, whereas our results are faithful to the Ground Truth.

最早核实视觉日期：2026-03-16；arXiv v1：2026-03-16；CVPR 2026 fixed official PDF

已查看固定PDF及目标图；以最早已核实匹配版本日期作时间代理，不声称实际作图日期。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Lei Zhang；137335；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[80, 71, 521, 289]，PDF点、左上角原点。

Figure 2. The framework of GDPO, which consists of two core stages: (a) advantage calculation and (b) policy optimization. Firstly, we
employ a pre-trained one-step Real-ISR model as the reference model to generate a group of diverse outputs by injecting different random
noises. Subsequently, we compute the advantage A for each sample by evaluating its reward with our designed attribute-aware reward
functions and converting these rewards into group-relative advantages. In the policy optimization stage, we feed these samples along with
noises into both the policy model and the reference ISR model, and update the parameters of the policy ISR model by minimizing the
proposed GDPO loss, steering it to favor generating high-reward samples.

最早核实视觉日期：2026-03-16；arXiv v1：2026-03-16；CVPR 2026 fixed official PDF

已查看固定PDF及目标图；以最早已核实匹配版本日期作时间代理，不声称实际作图日期。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Lei Zhang；137335；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[105, 70, 505, 250]，PDF点、左上角原点。

Figure 1. Noise regulates the diversity of generated samples. Different noise inputs yield both high-quality samples (e.g., noise1) and
low-quality ones (e.g., noise2 and noise3). After preference learning, the model produces more visually pleasing results.

最早核实视觉日期：2025-11-21；arXiv v1：2025-11-21；CVPR 2026 fixed official PDF

已核对固定会议PDF和arXiv v1的完整目标图；构图、文字、配色、元素和连线一致，v1为目前最早可核实的视觉版本。作者和机构已在PDF首页与公开引用主页交叉核对。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Ya Zhang；25506；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[59, 70, 555, 265]，PDF点、左上角原点。

Figure 3. Overall framework. We extend the visual stream into a frozen Prior Noise stream that preserves the pretrained diffusion prior
and a Control Noise stream that adaptively modulates noise based on the Fidelity Weight to restore fine details, forming a Noise-hybrid
Visual Stream. Leveraging the Fidelity Weight to adjust adversarial signals adaptively, Fidelity-aware Adversarial Training enhances
controllability and achieves one-step inference.

最早核实视觉日期：2025-11-21；arXiv v1：2025-11-21；CVPR 2026 fixed official PDF

已核对固定会议PDF和arXiv v1的完整目标图；构图、文字、配色、元素和连线一致，v1为目前最早可核实的视觉版本。作者和机构已在PDF首页与公开引用主页交叉核对。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Ya Zhang；25506；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[60, 70, 554, 180]，PDF点、左上角原点。

Figure 2. Effects of Noise on Fidelity and Controllability. Based on pretrained Qwen-Image, (a) shows the denoising results of ground-
truth (GT) under the same prompt and different levels of noise. (b) adopts low noise level and different prompts (with Chinese annotation).
(c) studies the effects of noise on LQ with the same prompt. High-noise inputs improve perceptual quality and controllability but reduce
fidelity, whereas low-noise inputs preserve original details yet fail to deliver enhanced super-resolution effects.

最早核实视觉日期：2026-03-25；arXiv v1：2026-03-25；ECCV 2026 fixed official PDF

已核对固定会议PDF和arXiv v1的完整目标图；构图、文字、配色、元素和连线一致，v1为目前最早可核实的视觉版本。作者和机构已在PDF首页与公开引用主页交叉核对。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Jing Liu；22235；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[135, 116, 478, 319]，PDF点、左上角原点。

Fig. 2: The proposed RefReward-SR framework. (a) MLLM Fine-tuning:
Fine-tuning via GRPO with format and LR-conditioned rank rewards from RefSR-
18K for human preference alignment. (b) Global-Local Crop Scoring: Extracting
representative local crops via RAM and Grounding DINO, then fusing multi-scale
MLLM scores via area-weighted averaging for comprehensive evaluation.

最早核实视觉日期：2026-03-25；arXiv v1：2026-03-25；ECCV 2026 fixed official PDF

已核对固定会议PDF和arXiv v1的完整目标图；构图、文字、配色、元素和连线一致，v1为目前最早可核实的视觉版本。作者和机构已在PDF首页与公开引用主页交叉核对。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Jing Liu；22235；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[135, 116, 478, 255]，PDF点、左上角原点。

Fig. 1: Overview of our proposed evaluation paradigm. (a) HR A preserves structures,
while HR B shows severe distortions. (b) RefReward-SR uses an MLLM for inter-
pretable, LR-conditioned semantic scoring. (c) Existing metrics wrongly favor the dis-
torted HR B, whereas ours aligns with human perception. (d) RefReward-SR uniquely
ensures semantic fidelity and preference alignment without GT dependency. The as-
terisk(*) indicates that FR metrics evaluate high-level semantics only passively via
low-level correspondence.

最早核实视觉日期：2026-04-18；arXiv v1：2025-10-27；CVPR 2026 fixed official PDF

已核查v1(2025-10-27)、v2(2025-11-06)、v3(2026-04-18)：早期图1(c)绿色概率路径及底部域图形位置不同，图2分数函数标签与公式的相对位置改变；v3与会议目标图像素一致。因此不回填v1日期，最早核实匹配版本为v3。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Bo Du；50901；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[110, 70, 502, 213]，PDF点、左上角原点。

Figure 2. A schematic of Residual Diffusion Bridge Models. RDBM utilizes a diffusion process guided by Doob’s h-transform towards an endpoint xT = µ free from stationary noise λϵ. Modulated by the residual component π = x0 − xT , the noise perturbation is selectively imposed on different regions with diverse degradation levels, thereby constructing probabilistic paths. Besides, it learns to reverse the process by matching the residual bridge score functions, facilitating an adaptive inversion from xT ∼ pprior(x) to x0 ∼ pdata(x).

最早核实视觉日期：2026-04-18；arXiv v1：2025-10-27；CVPR 2026 fixed official PDF

已核查v1(2025-10-27)、v2(2025-11-06)、v3(2026-04-18)：早期图1(c)绿色概率路径及底部域图形位置不同，图2分数函数标签与公式的相对位置改变；v3与会议目标图像素一致。因此不回填v1日期，最早核实匹配版本为v3。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Bo Du；50901；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[317, 295, 552, 543.7]，PDF点、左上角原点。

Figure 1. Typical diffusion processes. (a) Standard diffusion maps high-quality images to Gaussian noise domain. (b) Mean-reverting diffusion drives terminal state toward a low-quality domain with stationary noise. (c) Diffusion bridge establishes direct probabilistic transitions between known distributions. All inject noise globally, disrupting overall structures and constraining transitions. (d) In contrast, our RDBM selectively reconstructs degraded regions (e.g., the doll) while preserving intact areas (e.g., the background).

最早核实视觉日期：2026-03-19；arXiv v1：2026-03-19；CVPR 2026 fixed official PDF

已对照会议固定PDF与arXiv v1目标图，构图、文字和连线一致，导出像素一致；未发现更早公开版本。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Ying Fu；11391；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[70, 70, 522, 221]，PDF点、左上角原点。

Figure 2. The framework of our statistical-characteristic guided network. It enhances signals and suppresses noise in the spatial and
frequency domains using spatial deviation-guided weighting and frequency band-guided weighting, respectively.

最早核实视觉日期：2026-03-19；arXiv v1：2026-03-19；CVPR 2026 fixed official PDF

已对照会议固定PDF与arXiv v1目标图，构图、文字和连线一致，导出像素一致；未发现更早公开版本。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Ying Fu；11391；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[308, 70, 547, 124]，PDF点、左上角原点。

Figure 1. Visualization of the two kinds of statistical characteris-
tics that we used to guide the denoising process. (a) The spatial
deviation characteristic obtained by calculating the pixel standard
deviation (SD) within 3 × 3 windows. It reflects the fluctuation
conditions of different spatial positions in the image. (b) The fre-
quencybandcharacteristicobtainedbytheFastFourierTransform
(FFT). Signal and noise are distributed at different frequency band
positions and have different patterns.

最早核实视觉日期：2026-03-29；arXiv v1：2026-03-29；ECCV 2026 fixed official PDF

已查看会议版和arXiv v1：图内构图、图标、颜色、模块文字和连线一致；会议排版缩小并将框架图编号由2改为3，分别保存坐标，不将图号变化视作重绘。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Bowen Zhou；32977；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[132, 112, 484, 248]，PDF点、左上角原点。

Fig.3: Training pipeline of TIR-Agent. The process begins with SFT Data Gen-
eration, followed by Exploration-Driven Perturbation to enhance data diversity
via shuffled task sequences and tool substitutions. Following SFT, the agent undergoes
RL with a Multi-Dimensional Adaptive Reward mechanism, which dynamically
balances metric weights based on their EMA trends to optimize the restoration policy.

最早核实视觉日期：2026-03-29；arXiv v1：2026-03-29；ECCV 2026 fixed official PDF

已查看会议版和arXiv v1：图内构图、图标、颜色、模块文字和连线一致；会议排版缩小并将框架图编号由2改为3，分别保存坐标，不将图号变化视作重绘。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Bowen Zhou；32977；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[134, 113, 482, 332]，PDF点、左上角原点。

Fig.1: Comparison of restoration paradigms. (a) All-in-one model: Uses
a single network with specialized heads to inject various degradation features. (b)
Training-free Agent: Uses VLMs for static task planning, executing sub-tasks se-
quentially via exhaustive tool traversal. (c) TIR-Agent (Ours): A trainable agent
that dynamically decides the next sub-task based on current results and learns to di-
rectly select the optimal tool, avoiding redundant execution.

最早核实视觉日期：2025-12-07；arXiv v1：2025-12-07；CVPR 2026 fixed official PDF

已逐张查看会议版与arXiv v1：图1因作者行排版整体下移9.963 PDF点，设计/文本/布局一致；图2位置及设计一致。记录v1公开时间，未用会议录用时间。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Jian Zhang（张健）；21725；快照2026-10-21。

绘图工具：unknown。已读图注与相关正文；未找到该插图实际制作工具声明。

裁剪坐标：[58, 71, 554, 338]，PDF点、左上角原点。

Figure2. Illustrationofthearchitectureandtwo-stagetrainingframeworkofUARE.TwotransformerexpertsareusedtoprocessIQA
and restoration, respectively. Training stages include (1) a progressive, easy-to-hard schedule that moves from single-type to high-order
degradations. In this stage, only the restoration expert is trained to make UARE handle multiple degradations. (2) Unified fine-tuning of
the entire model to strengthen the IQA ability and align the IQA signals with restoration objectives through interleaved data.

最早核实视觉日期：2025-12-07；arXiv v1：2025-12-07；CVPR 2026 fixed official PDF

已逐张查看会议版与arXiv v1：图1因作者行排版整体下移9.963 PDF点，设计/文本/布局一致；图2位置及设计一致。记录v1公开时间，未用会议录用时间。

历史PDF · 下载引用证据快照 · 公开日期

作者引用：Jian Zhang（张健）；21725；快照2026-10-21。

绘图工具：unknown。已读图注与相关正文；未找到该插图实际制作工具声明。

裁剪坐标：[58, 237, 554, 528]，PDF点、左上角原点。

Figure 1. Showcase of UARE. It supports image quality assessment (image scoring and comparison), image restoration/enhancement
(super-resolution, dehazing, and low-light enhancement, etc.), and assessment-guided restoration and enhancement in one unified model.

最早核实视觉日期：2025-10-23；arXiv v1：2025-10-23；CVPR 2026 fixed official PDF

v1(2025-10-23)第4页图2已具有同一完整框架；会议版编号改为3并整体等比放大，模块、连线、配色、内容一致。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Shenghua Gao；29917；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[58, 70, 554, 253]，PDF点、左上角原点。

Figure 3. Overview of CUPID’s pipeline. From a single input image, CUPID first generates an occupancy and a UV cube in canonical
space. Then, a Perspective-n-Point (PnP) solver (i.e., Eq. 2) recovers the camera pose. Using this recovered camera pose, we extract pose-
aligned conditioning latents and visual features (i.e., Eq. 3), along with noisy structured latents, to generate the geometry and appearance
features, which will be decoded to the 3D Gaussian splats and mesh.

最早核实视觉日期：2025-11-24；arXiv v1：2025-10-23；CVPR 2026 fixed official PDF

v1第2页为大幅多样本结果拼图，尚无目标图1；v2(2025-11-24)第1页出现与会议版一致的联合物体/相机任务图，故按v2日期，不回填v1。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Shenghua Gao；29917；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[317, 228, 554, 338]，PDF点、左上角原点。

Figure 1. Generative 3D reconstruction from a single image.
Left: Input image. Middle: Our method recovers a canonical 3D
model (intrinsic properties such as shape and texture) along with
theexactcamerapose(extrinsicproperty)thatgeneratedtheinput.
Right: Re-rendering the 3D model from the estimated pose faith-
fully reproduces the input.

最早核实视觉日期：2025-12-11；arXiv v1：2025-12-11；CVPR 2026 fixed official PDF

已对照固定会议版本与arXiv v1，完整布局、标签、元素与连线匹配；v1为目前最早可核实的该视觉版本。已在PDF首页核对作者及单位。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Shubham Tulsiani；13507；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[58, 70, 555, 197]，PDF点、左上角原点。

Figure 2. E-RayZer Model & Training. E-RayZer first predicts camera poses and intrinsics for all images. Then it follows RayZer [25] to
split images into two sets. E-RayZer predicts explicit 3D Gaussians as scene representation from the reference views (I ), and renders the
ref
scene using self-predicted target-view (I ) cameras. Finally, E-RayZer is trained with self-supervised photometric losses on target views.
tgt

最早核实视觉日期：2025-12-11；arXiv v1：2025-12-11；CVPR 2026 fixed official PDF

已对照固定会议版本与arXiv v1，完整布局、标签、元素与连线匹配；v1为目前最早可核实的该视觉版本。已在PDF首页核对作者及单位。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Shubham Tulsiani；13507；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[58, 70, 296, 197]，PDF点、左上角原点。

Figure 3. Different Visual Overlaps under the Same Frame In-
terval. Two sequences from DL3DV [34] share the same frame in-
terval yet exhibit drastically different levels of visual overlap. Our
proposed semantic and geometric overlap metrics more accurately
capture the true difficulty (or camera motion) of each sequence.

最早核实视觉日期：2026-03-13；arXiv v1：2025-12-07；CVPR 2026 fixed official PDF

已查看v1(2025-12-07)图1第2页与图2第6页：图1重新布局、配色和标注；图2字体及标签排布改变。v2(2026-03-13)两图与会议版匹配（MAE 0），采用v2日期。两版均在S3，不将重绘风格回溯v1。历史对照另见quota_review/EM_v1p2.png与EM_v1_F.png。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Xiaokang Yang；50795；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[317, 70, 554, 336]，PDF点、左上角原点。

Figure 2. Overview of the pipeline of EMGauss. Given a vol-
ume EM dataset (top), the gray-shaded regions denote the miss-
ing intermediate slices to be interpolated, while the visible regions
represent the available anisotropic training data. We interpret the
low-resolution z -axis as a temporal dimension and employ a de-
formable 2D Gaussian splatting representation to model the ge-
ometric evolution of 2D structures along depth. During inference,
the model synthesizes the target slice by directly conditioning on
the relative temporal coordinate corresponding to the desired in-
terpolation position.

最早核实视觉日期：2026-03-13；arXiv v1：2025-12-07；CVPR 2026 fixed official PDF

已查看v1(2025-12-07)图1第2页与图2第6页：图1重新布局、配色和标注；图2字体及标签排布改变。v2(2026-03-13)两图与会议版匹配（MAE 0），采用v2日期。两版均在S3，不将重绘风格回溯v1。历史对照另见quota_review/EM_v1p2.png与EM_v1_F.png。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Xiaokang Yang；50795；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[58, 70, 298, 309]，PDF点、左上角原点。

Figure 1. Different paradigms of isotropic reconstruction from
anisotropic datasets. Existing approaches typically extract vol-
ume features from abundant xy-plane images. After stacking
anisotropic slices along the z-axis (top), they: (a) Video inter-
polation along the z-axis: models are trained along the x or y
direction, where adjacent frames are input to synthesize the inter-
mediate frame; during inference, the trained model is repurposed
to interpolate along the z-axis, generating intermediate slices cor-
responding to continuous temporal coordinates (τ ∈ [0, 1]). (b)
Image super-resolution on xz/yz planes: Since the xy-plane im-
ages are of much higher resolution, a low-resolution counterpart
can be generated either by manual downsampling [21] or by learn-
ing a degradation model [6], forming paired HR–LR supervision
for training; during inference, the resulting super-resolution model
is then applied along the z-axis to enhance xz/yz slices. Both
paradigms implicitly assume spatial isotropy. In contrast, our
method directly performs inference in continuous 3D space, with-
out relying on such isotropic assumptions.

最早核实视觉日期：2026-03-29；arXiv v1：2026-03-29；CVPR 2026 fixed official PDF

已对照固定会议版本与arXiv v1，完整布局、标签、元素与连线匹配；v1为目前最早可核实的该视觉版本。已在PDF首页核对作者及单位。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Krystian Mikolajczyk；49785；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[58, 70, 554, 196]，PDF点、左上角原点。

Figure 2. Training pipeline of NAS3R. Subscripts “C ” and “T ” denote context and target views, respectively. Unconstrained images are
patchified into visual tokens and concatenated with a learnable camera token for camera prediction. A masked decoder regulates cross-view
interactions and prevents target-to-context leakage. Refined context tokens are then processed by the Gaussian head to predict Gaussian
parameters, while a depth head estimates depth maps that are lifted into 3D Gaussian centers using the predicted context poses. The
predicted target poses are finally used to render novel views, providing photometric supervision for end-to-end training.

最早核实视觉日期：2026-03-29；arXiv v1：2026-03-29；CVPR 2026 fixed official PDF

已对照固定会议版本与arXiv v1，完整布局、标签、元素与连线匹配；v1为目前最早可核实的该视觉版本。已在PDF首页核对作者及单位。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Krystian Mikolajczyk；49785；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[58, 214, 554, 403]，PDF点、左上角原点。

Figure 1. NAS3R is a self-supervised framework that requires no ground-truth annotations and no pretrained priors during training, and
jointly infers 3D Gassian parameters, camera intrinsics and extrinsics and depth maps, also enabling high-quality novel view synthesis.

最早核实视觉日期：2025-11-26；arXiv v1：2025-11-26；CVPR 2026 fixed official PDF

已对照固定会议版本与arXiv v1，完整布局、标签、元素与连线匹配；v1为目前最早可核实的该视觉版本。已在PDF首页核对作者及单位。 作者Jiangmiao Pang个人主页oceanpang.github.io写明Shanghai AI Lab，且链接至与引用快照相同的Scholar ID ssSfKpAAAAAJ；Scholar单位字段较旧。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Jiangmiao Pang；26222；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[58, 70, 555, 274]，PDF点、左上角原点。

Figure 3. We present G²VLM, a unified model that integrates both a geometric perception expert for 3D reconstruction and a semantic
perception expert for multimodal understanding and spatial reasoning tasks. All tokens can do shared multi-modal self attention in each
transformer block.

最早核实视觉日期：2025-11-26；arXiv v1：2025-11-26；CVPR 2026 fixed official PDF

已对照固定会议版本与arXiv v1，完整布局、标签、元素与连线匹配；v1为目前最早可核实的该视觉版本。已在PDF首页核对作者及单位。 作者Jiangmiao Pang个人主页oceanpang.github.io写明Shanghai AI Lab，且链接至与引用快照相同的Scholar ID ssSfKpAAAAAJ；Scholar单位字段较旧。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Jiangmiao Pang；26222；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[58, 254, 550, 537]，PDF点、左上角原点。

Figure 1. We present G²VLM, a geometry grounded vision-language model proficient in both spatial 3D reconstruction and spatial
understanding tasks. For spatial reasoning questions, G²VLM can predict 3D geometry and employ interleaved reasoning for an answer.

最早核实视觉日期：2026-03-18；arXiv v1：2026-03-18；CVPR 2026 fixed official PDF

已核对固定会议PDF和arXiv v1的完整目标图；构图、文字、配色、元素和连线一致，v1为目前最早可核实的视觉版本。作者和机构已在PDF首页与公开引用主页交叉核对。 图1在会议版整体下移8.28 PDF点，平移对齐后像素平均差0.015/255；图2平均差小于0.001，未发现重绘。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Jingyi Yu；24450；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[58, 70, 555, 302]，PDF点、左上角原点。

Figure 2. Overview of the proposed PanoVGGT framework. Given one or multiple panoramic images, PanoVGGT extracts spherical
patch tokens using a ViT backbone, augments them via S O(3) geometric augmentation, and encodes them with shared and branch-adapted
positional embeddings. A multi-branch attention aggregator jointly reasons about local geometry, global structure, and camera motion.
The model outputs dense depth maps, camera poses (under a randomly selected global anchor), and local/global point maps for final 3D
reconstruction.

最早核实视觉日期：2026-03-18；arXiv v1：2026-03-18；CVPR 2026 fixed official PDF

已核对固定会议PDF和arXiv v1的完整目标图；构图、文字、配色、元素和连线一致，v1为目前最早可核实的视觉版本。作者和机构已在PDF首页与公开引用主页交叉核对。 图1在会议版整体下移8.28 PDF点，平移对齐后像素平均差0.015/255；图2平均差小于0.001，未发现重绘。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Jingyi Yu；24450；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[58, 211, 555, 479]，PDF点、左上角原点。

Figure 1. Overview of PanoVGGT and the PanoCity dataset. PanoVGGT predicts camera poses, dense depth, and consistent 3D point
clouds from unordered panoramas. Compared with perspective-based pipelines that split panoramas into pinhole views, it yields more
accurate 3D reconstructions. PanoCity contains over 120k outdoor panoramas under diverse urban scenes and weather conditions.

最早核实视觉日期：2026-04-17；arXiv v1：2026-04-17；CVPR 2026 fixed official PDF

已查看固定PDF及目标图；以最早已核实匹配版本日期作时间代理，不声称实际作图日期。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Nicu Sebe；59898；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[57, 72, 554, 146]，PDF点、左上角原点。

Figure 2. Framework of PoInit-of-View.

最早核实视觉日期：2026-04-17；arXiv v1：2026-04-17；CVPR 2026 fixed official PDF

已查看固定PDF及目标图；以最早已核实匹配版本日期作时间代理，不声称实际作图日期。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Nicu Sebe；59898；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[317, 232, 551, 390]，PDF点、左上角原点。

Figure1. OurPoInit-of-Viewmethodinjectsimperceptiblepertur-
bations into input views (top) to induce cross-view inconsistency
(seedetailsinFigure3)inthefoundationalStructure-from-Motion
(SfM)initializationratherthanthesingle-view-basedlossinexist-
ing attacks. The corrupted feature correspondences propagate to
inaccurate point triangulation, leading to erroneous camera poses
and severely degraded 3D reconstruction results (bottom). Com-
paredtocleanviews,ourpoisonedinputsyieldfarfewerregistered
images, and as a result, lower quality of rendered views.

最早核实视觉日期：2026-03-21；arXiv v1：2026-03-21；ECCV 2026 fixed official PDF

已核对固定会议PDF和arXiv v1的完整目标图；构图、文字、配色、元素和连线一致，v1为目前最早可核实的视觉版本。作者和机构已在PDF首页与公开引用主页交叉核对。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Yan Yan；13225；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[135, 116, 478, 262]，PDF点、左上角原点。

Fig. 4: Method Overview. Our method performs streaming 3D reconstruction via
dual-branch inference. At time step t, the main branch predicts depth and pose from
state s using both image and RayMap features. The predicted pose Tˆ is remapped
t−1 t
into RayMap and encoded into tokens r′ , then queried against the frozen state s
t t−1
by the RayMap branch to obtain a static-biased prediction. The depth difference
between branches is pro jected via image-state attention to form staticness weights α
t
that suppress dynamic regions during state update (s = s + α ⊙ ∆s ). This dual-
t t−1 t t
branch scheme identifies and suppresses dynamic regions at inference time.

最早核实视觉日期：2026-03-21；arXiv v1：2026-03-21；ECCV 2026 fixed official PDF

已核对固定会议PDF和arXiv v1的完整目标图；构图、文字、配色、元素和连线一致，v1为目前最早可核实的视觉版本。作者和机构已在PDF首页与公开引用主页交叉核对。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Yan Yan；13225；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[135, 116, 478, 227]，PDF点、左上角原点。

Fig. 2: RayMap Static Bias. Left: The main prediction and RayMap prediction pro-
duce differing depth estimates in dynamic regions, suggesting a static bias in RayMap
predictions. Right: The depth discrepancy correlates with ground-truth dynamic ratio
across multiple datasets, suggesting a consistent signal for dynamic identification.

最早核实视觉日期：2026-04-02；arXiv v1：2026-01-21；CVPR 2026 fixed official open-access version

核查v1(2026-01-21)、v2(2026-02-02)、v3(2026-03-13)、v4(2026-04-02)及会议固定版；v1-v3使用Mapping Field标签，v4改为RGB CRF Model/Event CRF Model并与会议目标图匹配。不能套用v1日期。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Jia Li；12248；快照2026-10-21。

绘图工具：unknown。已读图注、引言和方法段落；未获得插图制作工具声明。

裁剪坐标：[57, 70, 555, 250]，PDF点、左上角原点。

Figure 3. The data generation process of event and RGB cameras (upper parts) and the pipeline of See-NeRF (lower parts) are shown in the
figure. We use NeRF to represent the actual radiance of an extreme lighting scene in the HDR domain. The volume rendering simulates the
raw scene radiance rays hitting the sensor and obtaining the raw sensor values. The predicted LDR images and events are generated with our
proposed sensor-physics grounded RGB and event CRF models (details are shown in the right part of the figure). The image loss and event
loss are employed to jointly supervise the optimization of the two CRF models and the NeRF network.

最早核实视觉日期：2026-04-02；arXiv v1：2026-01-21；CVPR 2026 fixed official open-access version

核查v1(2026-01-21)、v2(2026-02-02)、v3(2026-03-13)、v4(2026-04-02)及会议固定版；v1-v3使用Mapping Field标签，v4改为RGB CRF Model/Event CRF Model并与会议目标图匹配。不能套用v1日期。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Jia Li；12248；快照2026-10-21。

绘图工具：unknown。已读图注、引言和方法段落；未获得插图制作工具声明。

裁剪坐标：[316, 255, 555, 558]，PDF点、左上角原点。

Figure 1. Events corresponding to a blurry LDR image contain
spatial difference (High Dynamic Range) and temporal difference
(High Temporal Resolution) information of the scene radiance,
making it an ideal data for enhancing HDR sharp 3D scene recon-
struction. See-NeRF models the physical imaging process with
proposed RGB and event CRF models to bridge the sensor-physics
discrepancy, leveraging events to infer scene dynamics and achiev-
ing sharper and better HDR novel view synthesis results compared
to the physical unaligned method EvHDR-NeRF [5].

最早核实视觉日期：2025-12-01；arXiv v1：2025-12-01；CVPR 2026 fixed official PDF

固定会议PDF与arXiv v1目标图按对应坐标逐图比对；像素均差为0；arXiv v1与已知会议投稿均在指定阶段。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Matthias Zwicker；17011；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[57, 71, 555, 225]，PDF点、左上角原点。

Figure 2. Overview of our SplatSuRe framework. A high-resolution (HR) 3D Gaussian Splatting (3DGS) model is trained using low-
resolution (LR) and super-resolution (SR) inputs. We first train a 3DGS model on LR inputs to identify undersampled regions and render
per-view weight maps that indicate where SR is needed. During training of the HR 3DGS model, the images produced by the frozen
single-image super-resolution (SISR) model are spatially weighted by these maps to form the SR loss L . A complementary LR loss
SR
L compares the downsampled HR render against the original LR ground truth to provide consistent supervision across the entire image.
LR

最早核实视觉日期：2025-12-01；arXiv v1：2025-12-01；CVPR 2026 fixed official PDF

固定会议PDF与arXiv v1目标图按对应坐标逐图比对；像素均差为0；arXiv v1与已知会议投稿均在指定阶段。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Matthias Zwicker；17011；快照2026-10-21。

绘图工具：unknown。未发现可核实的绘图工具披露；不根据外观猜测。

裁剪坐标：[337, 70, 531, 249]，PDF点、左上角原点。

Figure 3. Disparity in high-frequency ground truth informa-
tion across different views. Low-resolution ground truth from
near cameras provides high-resolution information for rendering
distant views, reducing the need for additional generated detail
in those views. Conversely, super-resolution is needed in views
where no other camera provides higher-resolution information.

最早核实视觉日期：2026-05-10；arXiv v1：2026-05-10；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

v1多尺度可变深度结构匹配，正文排版位置不同

历史PDF · 公开日期 · 论文入口

作者引用：按用户授权免除；未知值没有记为0。

绘图工具：unknown。未发现针对所选M/F图的具体绘图工具披露；一般写作/代码AI声明不作绘图工具推断。

裁剪坐标：[144, 79, 477, 222]，PDF点、左上角原点。

Figure 1: (a) Overall pipeline of our any-depth object detection network, following the standard
backbone–neck–head design common to modern detectors such as RT-DETR and YOLO, operating
on multi-scale features. Each stage in both the backbone and neck is divided into an essential path
(solid blocks) and a skippable refinement path (dashed blocks). At inference time, any combination
of essential and full paths across stages yields a valid sub-network, enabling a continuous range of
accuracy–efficiency trade-offs from a single set of weights. (b) A residual-block stage, as used in
RT-DETR. (c) A CSP-style stage, as used in the YOLO family. In both stage types, the essential and
full paths are jointly trained via self-distillation to produce aligned feature representations at each
stage boundary, so that downstream stages can process the output of either path reliably.

最早核实视觉日期：2026-05-10；arXiv v1：2026-05-10；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

v1目标指派冲突示意匹配

历史PDF · 公开日期 · 论文入口

作者引用：按用户授权免除；未知值没有记为0。

绘图工具：unknown。未发现针对所选M/F图的具体绘图工具披露；一般写作/代码AI声明不作绘图工具推断。

裁剪坐标：[124, 79, 491, 203]，PDF点、左上角原点。

Figure 2: Conflicting target assignment in any-depth detectors. (a) In AnyDepth-DETR, the Hungarian
matcher can assign different queries to the same ground-truth objects across the two execution paths.
We reorder the super-net’s foreground queries to align with the base-net’s target-matched query
ordering before applying KD. (b) In AnyDepth-YOLO, the Task-Aligned Assigner (TAL) can assign
the super-net and base-net outputs to different anchors for the same ground-truth object. We exclude
conflicting anchor assignments (in red boxes) from the foreground distillation set.

最早核实视觉日期：2026-10-03；arXiv v1：未匹配；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

完整arXiv标题检索未找到匹配；补充标题检索未找到可确认的更早同稿。证据为公开投稿时间代理。

公开日期 · 论文入口

作者引用：按用户授权免除；未知值没有记为0。

绘图工具：unknown。未发现针对所选M/F图的具体绘图工具披露；一般写作/代码AI声明不作绘图工具推断。

裁剪坐标：[106, 79, 507, 272]，PDF点、左上角原点。

Figure 2: SCLE only modifies the pre-decoder selection stage of the language-guided detector. Left: detection pipeline, where SCLE-driven query selection replaces text-score ranking before the HED decoder. Upper right: class-wise foreground prototypes and one shared background prototype, updated by exponential moving average from labeled support samples. Lower right: the prototype bank produces a class-agnostic evidence map. Candidates are scored using interior purity P_i, ring evidence R_i, and contrast C_i. A quota divides the Top-900 budget into semantic, local-evidence and joint channels. Only the indices fed into the decoder are changed.

最早核实视觉日期：2026-10-03；arXiv v1：未匹配；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

完整arXiv标题检索未找到匹配；补充标题检索未找到可确认的更早同稿。证据为公开投稿时间代理。

公开日期 · 论文入口

作者引用：按用户授权免除；未知值没有记为0。

绘图工具：unknown。未发现针对所选M/F图的具体绘图工具披露；一般写作/代码AI声明不作绘图工具推断。

裁剪坐标：[106, 79, 507, 252]，PDF点、左上角原点。

Figure 1: (a) A brief illustration of the current Grounding-DINO-based works. (b) We find that
compared with the ground-truth bounding box (red), for target domains, the top-ranked candidate
boxes (yellow) are less likely to tightly bound the objects, where the better candidate boxes (cyan)
may be ranked outside the Top-900 quota. Unlike current work, we identify the selected candidate
boxes as another bottleneck for the CD-FSOD task, which we aim to address in this paper.

最早核实视觉日期：2026-10-03；arXiv v1：未匹配；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

完整arXiv标题检索未找到匹配；补充标题检索未找到可确认的更早同稿。证据为公开投稿时间代理。

公开日期 · 论文入口

作者引用：按用户授权免除；未知值没有记为0。

绘图工具：unknown。未发现针对所选M/F图的具体绘图工具披露；一般写作/代码AI声明不作绘图工具推断。

裁剪坐标：[128, 79, 488, 219]，PDF点、左上角原点。

Figure 2: Overview of the proposed DCHA framework. (a) DCHA constructs category-specific
LoRA by composing vectors from a shared bank and averages the resulting category updates to form
the task-level adaptation for the VLF layers. (b) LCAR transforms category-specific routing logits
into sparse composition weights using a learnable sparsity coefficient, enabling category-dependent
vector selection. (c) HDC preserves historical adaptation knowledge by constraining deviations at
the bank-vector, category-update, and task-update levels.

最早核实视觉日期：2026-10-03；arXiv v1：未匹配；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

完整arXiv标题检索未找到匹配；补充标题检索未找到可确认的更早同稿。证据为公开投稿时间代理。

公开日期 · 论文入口

作者引用：按用户授权免除；未知值没有记为0。

绘图工具：unknown。未发现针对所选M/F图的具体绘图工具披露；一般写作/代码AI声明不作绘图工具推断。

裁剪坐标：[118, 79, 488, 178]，PDF点、左上角原点。

Figure 1: Motivation of DCHA. (a) Existing category-level methods rigidly assign fixed experts
based on category, whereas DCHA dynamically selects and composes fine-grained parameter vec-
tors according to category demands. (b) Category-specific updates produced by existing category-
level methods exhibit uniformly high pairwise similarities within the same task. (c) Shared bank
parameter deviations progressively accumulate at the vector, category, and task levels during incre-
mental training, resulting in hierarchical drift.

最早核实视觉日期：2026-10-03；arXiv v1：未匹配；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

完整arXiv标题检索未找到匹配；补充标题检索未找到可确认的更早同稿。证据为公开投稿时间代理。

公开日期 · 论文入口

作者引用：按用户授权免除；未知值没有记为0。

绘图工具：unknown。作者声明AI编辑辅助分析及可视化代码；未披露所选M/F绘制工具。

裁剪坐标：[106, 79, 507, 311]，PDF点、左上角原点。

Figure 2: Overview of DeCo. Independent backbones extract multimodal, multiscale features, which
are projected into a shared density space to estimate modality reliability and construct density con-
sensus. The resulting consensus guides multiscale feature fusion for detection and is simultaneously
decoded into a density map for density supervision.

最早核实视觉日期：2026-10-03；arXiv v1：未匹配；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

完整arXiv标题检索未找到匹配；补充标题检索未找到可确认的更早同稿。证据为公开投稿时间代理。

公开日期 · 论文入口

作者引用：按用户授权免除；未知值没有记为0。

绘图工具：unknown。作者声明AI编辑辅助分析及可视化代码；未披露所选M/F绘制工具。

裁剪坐标：[106, 79, 507, 261]，PDF点、左上角原点。

Figure 1: Motivation for density consensus learning. OPT and SAR observations provide heteroge-
neous and spatially varying evidence for the same underlying ship distribution. We establish their
agreement in a shared density space to preserve mutually supported target responses while suppress-
ing modality-specific interference. Different Gaussian bandwidths in (d) illustrate the dependence
of conventional pixel-wise density targets on predefined kernel shapes.

最早核实视觉日期：2026-10-03；arXiv v1：未匹配；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

完整arXiv标题检索未找到匹配；补充标题检索未找到可确认的更早同稿。证据为公开投稿时间代理。

公开日期 · 论文入口

作者引用：按用户授权免除；未知值没有记为0。

绘图工具：unknown。未发现针对所选M/F图的具体绘图工具披露；一般写作/代码AI声明不作绘图工具推断。

裁剪坐标：[119, 77, 494, 263]，PDF点、左上角原点。

Figure 3: Overview of InstaPoint. IDC initializes instance-wise assignments to SAM or watershed
using class-calibrated point spacing. ISA then makes a one-time selection of the source with the
lower GWD-based size loss relative to the detector’s current prediction.

最早核实视觉日期：2026-10-03；arXiv v1：未匹配；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

完整arXiv标题检索未找到匹配；补充标题检索未找到可确认的更早同稿。证据为公开投稿时间代理。

公开日期 · 论文入口

作者引用：按用户授权免除；未知值没有记为0。

绘图工具：unknown。未发现针对所选M/F图的具体绘图工具披露；一般写作/代码AI声明不作绘图工具推断。

裁剪坐标：[123, 79, 486, 210]，PDF点、左上角原点。

Figure 1: Qualitative comparison of supervision assignment. Point2RBox-v3 (Zhang et al., 2026)
uses a shared patch-wise supervision source, while our Instance Density Calibration (IDC) and
Instance-wise Supervision Assignment (ISA) perform instance-wise source assignment between
SAM (blue) and watershed (green) supervision. Yellow boxes denote the RBoxes and white dots
denote point annotations.

最早核实视觉日期：2026-10-03；arXiv v1：未匹配；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

完整arXiv标题检索未找到匹配；补充标题检索未找到可确认的更早同稿。证据为公开投稿时间代理。

公开日期 · 论文入口

作者引用：按用户授权免除；未知值没有记为0。

绘图工具：unknown。未发现针对所选M/F图的具体绘图工具披露；一般写作/代码AI声明不作绘图工具推断。

裁剪坐标：[106, 79, 507, 252]，PDF点、左上角原点。

Figure 3: An illustration of our proposed IACCL

最早核实视觉日期：2026-10-03；arXiv v1：未匹配；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

完整arXiv标题检索未找到匹配；补充标题检索未找到可确认的更早同稿。证据为公开投稿时间代理。

公开日期 · 论文入口

作者引用：按用户授权免除；未知值没有记为0。

绘图工具：unknown。未发现针对所选M/F图的具体绘图工具披露；一般写作/代码AI声明不作绘图工具推断。

裁剪坐标：[102, 79, 507, 213]，PDF点、左上角原点。

Figure 1: Long-tailed category distributions in the DOTA v1.0 and DOTA v1.5 training sets.

最早核实视觉日期：2026-10-03；arXiv v1：未匹配；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

完整arXiv标题检索未找到匹配；补充标题检索未找到可确认的更早同稿。证据为公开投稿时间代理。

公开日期 · 论文入口

作者引用：按用户授权免除；未知值没有记为0。

绘图工具：generative_AI_figure_preparation_unspecified_tool。作者声明生成式AI用于figure preparation；工具名及逐图对应未披露。

裁剪坐标：[110, 169, 507, 348]，PDF点、左上角原点。

Figure 2: Persistent support evidence and SSCA-centered calibration in PULSE. In Stage I, HED
is fine-tuned on the target support set, and the resulting encoder snapshot is frozen to construct the
evidence bank E. In Stage II, the query detector is initialized with the Stage I weights and produces
query features Q, base class logits S, and predicted boxes B. SSCA matches Q with the cached
evidence E to obtain support agreement s and refines the base logits as S′ = S + G ⊙ ∆, while the
box coordinates bypass score calibration and remain unchanged. The evidence bank remains fixed
throughout Stage II training and inference. Flames indicate trainable parameters during training,
whereas snowflakes indicate frozen modules or fixed evidence. All model parameters are fixed during
inference. Test images from the DeepFish dataset (Fu et al., 2024) are adopted for illustration and do
not correspond to the sampled adaptation episode; the displayed output box is schematic rather than
an actual model prediction.

最早核实视觉日期：2026-10-03；arXiv v1：未匹配；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

完整arXiv标题检索未找到匹配；补充标题检索未找到可确认的更早同稿。证据为公开投稿时间代理。

公开日期 · 论文入口

作者引用：按用户授权免除；未知值没有记为0。

绘图工具：generative_AI_figure_preparation_unspecified_tool。作者声明生成式AI用于figure preparation；工具名及逐图对应未披露。

裁剪坐标：[110, 79, 501, 301]，PDF点、左上角原点。

Figure 1: Target-support use in FT-FSOD (Yu et al., 2026) and PULSE. (a) FT-FSOD uses support
annotations for target-domain fine-tuning and subsequently scores query predictions without explicitly
consulting the support examples. (b) PULSE retains cached support evidence and uses SSCA to
refine classification scores according to support–query agreement. Box tags denote predicted classes,
and bar lengths illustrate schematic confidence. The dashed candidate illustrates score suppression at
the same location. Test images from DeepFish (Fu et al., 2024) are used for illustration; the support
examples, bounding boxes, and score changes are schematic.

最早核实视觉日期：2026-10-03；arXiv v1：未匹配；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

完整arXiv标题检索未找到匹配；补充标题检索未找到可确认的更早同稿。证据为公开投稿时间代理。

公开日期 · 论文入口

作者引用：按用户授权免除；未知值没有记为0。

绘图工具：unknown。未发现针对所选M/F图的具体绘图工具披露；一般写作/代码AI声明不作绘图工具推断。

裁剪坐标：[106, 79, 507, 222]，PDF点、左上角原点。

Figure 1: POISE overview. The support set is encoded by a frozen DINOv2 encoder into a raw memory
bank and per-class prototype maps. TPR retrieves an adjacency set of hard negatives, renews each prototype
by a bounded one-step update, and writes the refined prototype back. UPN proposals are scored by cosine
classification against the refined prototypes; suspect screening flags nested same-class pairs and passes only
those to OAD, which separates whole objects from fragments.

最早核实视觉日期：2026-10-03；arXiv v1：未匹配；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

完整arXiv标题检索未找到匹配；补充标题检索未找到可确认的更早同稿。证据为公开投稿时间代理。

公开日期 · 论文入口

作者引用：按用户授权免除；未知值没有记为0。

绘图工具：unknown。未发现针对所选M/F图的具体绘图工具披露；一般写作/代码AI声明不作绘图工具推断。

裁剪坐标：[106, 79, 507, 179]，PDF点、左上角原点。

Figure 2: Why over-segmentation defeats confidence-based suppression in the few-shot regime. (a) Real
scenes contain multiple overlapping same-class instances, so class-agnostic proposal generators fragment ob-
jects into nested, mutually overlapping proposals that simple merging cannot consolidate. (b) The smaller
member of a nested pair is occasionally a genuine, distinct object, so the covered box cannot simply be dis-
carded. (c) Few-shot support sets are frequently biased: a handful of examples may emphasise part-views and
fail to represent the class as a whole. (d) Under such bias, the prototype shifts towards part-views, and an
over-segmentation fragment can attain higher classification confidence than the correct whole-object proposal.

最早核实视觉日期：2026-10-03；arXiv v1：未匹配；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

完整arXiv标题检索未找到匹配；补充标题检索未找到可确认的更早同稿。证据为公开投稿时间代理。

公开日期 · 论文入口

作者引用：按用户授权免除；未知值没有记为0。

绘图工具：unknown。未发现针对所选M/F图的具体绘图工具披露；一般写作/代码AI声明不作绘图工具推断。

裁剪坐标：[106, 564, 507, 713]，PDF点、左上角原点。

Figure 1: Schematic Overview of the CABE Framework Pipeline

最早核实视觉日期：2026-10-03；arXiv v1：未匹配；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

完整arXiv标题检索未找到匹配；补充标题检索未找到可确认的更早同稿。证据为公开投稿时间代理。

公开日期 · 论文入口

作者引用：按用户授权免除；未知值没有记为0。

绘图工具：unknown。未发现针对所选M/F图的具体绘图工具披露；一般写作/代码AI声明不作绘图工具推断。

裁剪坐标：[163, 79, 450, 290]，PDF点、左上角原点。

Figure 2: The relationship between Expansion pixels and the empirical FNR across different images.

最早核实视觉日期：2026-10-03；arXiv v1：未匹配；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

完整arXiv标题检索未找到匹配；补充标题检索未找到可确认的更早同稿。证据为公开投稿时间代理。

公开日期 · 论文入口

作者引用：按用户授权免除；未知值没有记为0。

绘图工具：unknown。未发现针对所选M/F图的具体绘图工具披露；一般写作/代码AI声明不作绘图工具推断。

裁剪坐标：[106, 79, 507, 257]，PDF点、左上角原点。

Figure 2: Overview of the UniOSE model. (a) Semantics-Enriched Object Namer combines
Deformable region reader layer
detector features with region-level VLM semantics to construct queries z and generates category
i
names with a generative lang F uage model. (b) Semantics-Guided Object Dx2etector predicts candi-
q Deformable
date boxes and combines query–targ+et caLteaygeroNroyrmmatchFiFnNg cos+ts wiLtahyesrNpoarmtial costs for training-time
Cross-Attention
supervision assignment. (c) Semantically Conditioned Mask Decoder injects z into the first mask
i
token and combines it with image features and predicted boxes to generate masks.

最早核实视觉日期：2026-10-03；arXiv v1：未匹配；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

完整arXiv标题检索未找到匹配；补充标题检索未找到可确认的更早同稿。证据为公开投稿时间代理。

公开日期 · 论文入口

作者引用：按用户授权免除；未知值没有记为0。

绘图工具：unknown。未发现针对所选M/F图的具体绘图工具披露；一般写作/代码AI声明不作绘图工具推断。

裁剪坐标：[106, 79, 507, 191]，PDF点、左上角原点。

Figure 1: Object semantics for open-ended naming and spatial prediction. (a) Spatial prediction produces object queries for LLM-based naming. (b) UniOSE enriches these queries with image-derived regional semantics and uses them for training-time detection matching and mask conditioning. Semantic guidance originates from the enriched query before LLM-based naming. (c) Comparison of box recall@0.75 and naming accuracy. Swin and SAM denote Open-Det Swin-T and its fully fine-tuned SAM2-T backbone variant, respectively. The protocol, metric definitions, and source results are provided in Appendix A.1.

最早核实视觉日期：2026-10-03；arXiv v1：未匹配；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

完整arXiv标题检索未找到匹配；补充标题检索未找到可确认的更早同稿。证据为公开投稿时间代理。

公开日期 · 论文入口

作者引用：按用户授权免除；未知值没有记为0。

绘图工具：unknown。未发现针对所选M/F图的具体绘图工具披露；一般写作/代码AI声明不作绘图工具推断。

裁剪坐标：[106, 79, 507, 133]，PDF点、左上角原点。

Figure 2: A-MoM integration with a Transformer. A single Transformer with shared weights
processes all modalities stacked along the batch dimension; A-MoM fuses them between layers,
independently at every token. A learned aggregation merges m tokens into 1 token at the stage output.

最早核实视觉日期：2026-10-03；arXiv v1：未匹配；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

完整arXiv标题检索未找到匹配；补充标题检索未找到可确认的更早同稿。证据为公开投稿时间代理。

公开日期 · 论文入口

作者引用：按用户授权免除；未知值没有记为0。

绘图工具：unknown。未发现针对所选M/F图的具体绘图工具披露；一般写作/代码AI声明不作绘图工具推断。

裁剪坐标：[106, 79, 507, 143]，PDF点、左上角原点。

Figure 1: Token-level motivation. Sensor advantages are local, so the fusion decision belongs at the
token level, illustrated by 1 case each for RGB, events, and depth.

最早核实视觉日期：2026-10-03；arXiv v1：未匹配；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

完整arXiv标题检索未找到匹配；补充标题检索未找到可确认的更早同稿。证据为公开投稿时间代理。

公开日期 · 论文入口

作者引用：按用户授权免除；未知值没有记为0。

绘图工具：unknown。未发现针对所选M/F图的具体绘图工具披露；一般写作/代码AI声明不作绘图工具推断。

裁剪坐标：[106, 79, 507, 293]，PDF点、左上角原点。

Figure 3: Overview of Classwise State Routing. Top: A labeled calibration set estimates down-
stream class-by-state reliability, selects a compact shared state pool, and fixes class-conditioned
routing weights. Bottom: For a new image, the frozen backbone extracts selected-state responses,
which are fused and rescaled before unchanged backend-native decoding.

最早核实视觉日期：2026-10-03；arXiv v1：未匹配；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

完整arXiv标题检索未找到匹配；补充标题检索未找到可确认的更早同稿。证据为公开投稿时间代理。

公开日期 · 论文入口

作者引用：按用户授权免除；未知值没有记为0。

绘图工具：unknown。未发现针对所选M/F图的具体绘图工具披露；一般写作/代码AI声明不作绘图工具推断。

裁剪坐标：[139, 80, 464, 372]，PDF点、左上角原点。

Figure 2: (a) Segmentation mIoU over 1,200 layer–timestep states; (b) Three-dimensional view
of the same performance landscape. (c) Classwise reliability over the selected states; stars mark
classwise maxima and red marks the strongest shared state. (d) Class IoU of the strongest shared
state and each class’s preferred state.

最早核实视觉日期：2026-10-03；arXiv v1：未匹配；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

完整arXiv标题检索未找到匹配；补充标题检索未找到可确认的更早同稿。证据为公开投稿时间代理。

公开日期 · 论文入口

作者引用：按用户授权免除；未知值没有记为0。

绘图工具：unknown。未发现针对所选M/F图的具体绘图工具披露；一般写作/代码AI声明不作绘图工具推断。

裁剪坐标：[106, 79, 507, 445]，PDF点、左上角原点。

Figure 2: DevoSeg’s decision process. The Router chooses policies before execution; the Compara-
tor selects among their masks and the independent Direct mask afterward. Completed outcomes
provide evidence for subsequent knowledge updates.

最早核实视觉日期：2026-10-03；arXiv v1：未匹配；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

完整arXiv标题检索未找到匹配；补充标题检索未找到可确认的更早同稿。证据为公开投稿时间代理。

公开日期 · 论文入口

作者引用：按用户授权免除；未知值没有记为0。

绘图工具：unknown。未发现针对所选M/F图的具体绘图工具披露；一般写作/代码AI声明不作绘图工具推断。

裁剪坐标：[106, 79, 507, 298]，PDF点、左上角原点。

Figure 1: Direct segmentation and a fixed tool pipeline follow predetermined paths. DevoSeg routes
policies per image, compares their masks, and uses completed cases to update its decision state.

最早核实视觉日期：2026-10-03；arXiv v1：未匹配；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

完整arXiv标题检索未找到匹配；补充标题检索未找到可确认的更早同稿。证据为公开投稿时间代理。

公开日期 · 论文入口

作者引用：按用户授权免除；未知值没有记为0。

绘图工具：unknown。未发现针对所选M/F图的具体绘图工具披露；一般写作/代码AI声明不作绘图工具推断。

裁剪坐标：[107, 80, 506, 263]，PDF点、左上角原点。

Figure 2: Overview of HIVE. A frozen visual encoder processes a global view and multiple overlap-
ping local views. HVR refines local features through local-to-global-to-local interaction, producing
holistic local representations {FH }R . IVE iteratively evolves an image-specific vocabulary set
Lr r=1
V∗ under visual guidance. Finally, the local predictions derived from {ZV∗}R are refined and
Lr r=1
fused to produce the final segmentation Y˜︁ V∗.

最早核实视觉日期：2026-10-03；arXiv v1：未匹配；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

完整arXiv标题检索未找到匹配；补充标题检索未找到可确认的更早同稿。证据为公开投稿时间代理。

公开日期 · 论文入口

作者引用：按用户授权免除；未知值没有记为0。

绘图工具：unknown。未发现针对所选M/F图的具体绘图工具披露；一般写作/代码AI声明不作绘图工具推断。

裁剪坐标：[126, 216, 486, 322]，PDF点、左上角原点。

Figure 1: Paradigm and performance comparison between conventional approach and our HIVE.

最早核实视觉日期：2026-10-03；arXiv v1：未匹配；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

完整arXiv标题检索未找到匹配；补充标题检索未找到可确认的更早同稿。证据为公开投稿时间代理。

公开日期 · 论文入口

作者引用：按用户授权免除；未知值没有记为0。

绘图工具：unknown。未发现针对所选M/F图的具体绘图工具披露；一般写作/代码AI声明不作绘图工具推断。

裁剪坐标：[107, 80, 506, 307]，PDF点、左上角原点。

Figure 3: Overview of AKI. (a) Three-level knowledge reorganization. (b) RCA constructs compact
response support and enforces response consistency. (c) RPA discovers reusable semantic directions
for selective alignment. (d) DEP suppresses current old–new distribution overlap.

最早核实视觉日期：2026-10-03；arXiv v1：未匹配；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

完整arXiv标题检索未找到匹配；补充标题检索未找到可确认的更早同稿。证据为公开投稿时间代理。

公开日期 · 论文入口

作者引用：按用户授权免除；未知值没有记为0。

绘图工具：unknown。未发现针对所选M/F图的具体绘图工具披露；一般写作/代码AI声明不作绘图工具推断。

裁剪坐标：[107, 80, 506, 261.4]，PDF点、左上角原点。

Figure 1: Motivation of AKI. (a) Under fixed model capacity, inherited constraints can restrict the representation reorganization required by current supervision, increasing old–new distribution overlap. (b) Cross-step optimization analysis shows that adjacent incremental objectives generally correspond to different stationary states, θ*_(t−1) ≠ θ*_t, while inheritance constraints can restrict current-task-aligned adaptation. (c) Empirical analysis reveals cross-level coupling among response, feature, and distribution variations, motivating coordinated knowledge reorganization across three levels.

最早核实视觉日期：2026-10-03；arXiv v1：未匹配；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

完整arXiv标题检索未找到匹配；补充标题检索未找到可确认的更早同稿。证据为公开投稿时间代理。

公开日期 · 论文入口

作者引用：按用户授权免除；未知值没有记为0。

绘图工具：unknown; authors_deny_generative_AI_assistance。第9页AI use statement明确表示未使用生成式AI辅助；具体绘图软件未知。

裁剪坐标：[107, 80, 505, 268]，PDF点、左上角原点。

Figure 2: Overview of ReTax-Seg. The upper panel summarizes the training pipeline. The visual
backbone and forward model are learnable. CLIP and SAM remain frozen. Boundary quality scores
the segmentation outputs. The lower panel details the multiscale visual backbone, the frozen Mobile-
SAM geometry branch, representative semantic atoms, and the frozen CLIP text-prior construction.
The symbol θ∗ denotes the weights associated with dataset d. The star marks the selected trained
d
checkpoint. For example, θ∗ represents the Cityscapes-specific weights. Snowflakes and flames
City
indicate frozen and learnable components, respectively.

最早核实视觉日期：2026-10-03；arXiv v1：未匹配；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

完整arXiv标题检索未找到匹配；补充标题检索未找到可确认的更早同稿。证据为公开投稿时间代理。

公开日期 · 论文入口

作者引用：按用户授权免除；未知值没有记为0。

绘图工具：unknown; authors_deny_generative_AI_assistance。第9页AI use statement明确表示未使用生成式AI辅助；具体绘图软件未知。

裁剪坐标：[107, 80, 505, 184]，PDF点、左上角原点。

Figure 1: Three unresolved challenges in taxonomy-aware segmentation. Left: identical visual
concepts can be split, merged, or omitted by different annotation policies. Center: a coarse label such
as vehicle is compatible with several semantic atoms, but the correct relation changes with local visual
evidence, which a class name alone cannot observe. Right: forward-only supervision can preserve an
apparently correct coarse mask even when distinct atom channels become indistinguishable. A useful
compiler must therefore be local, visually grounded, and semantically recoverable.

最早核实视觉日期：2026-10-03；arXiv v1：未匹配；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

完整arXiv标题检索未找到匹配；补充标题检索未找到可确认的更早同稿。证据为公开投稿时间代理。

公开日期 · 论文入口

作者引用：按用户授权免除；未知值没有记为0。

绘图工具：unknown。未发现针对所选M/F图的具体绘图工具披露；一般写作/代码AI声明不作绘图工具推断。

裁剪坐标：[107, 80, 506, 235]，PDF点、左上角原点。

Figure 2: Overview of the proposed UCMIT framework. At each scale, LCIT reconstructs RGB-
aligned thermal evidence from nearby thermal sources. REF then arbitrates between co-located and
transported evidence and regulates its residual injection into the RGB stream. The resulting multi-
scale representations are progressively aggregated for task-specific dense prediction.

最早核实视觉日期：2026-10-03；arXiv v1：未匹配；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

完整arXiv标题检索未找到匹配；补充标题检索未找到可确认的更早同稿。证据为公开投稿时间代理。

公开日期 · 论文入口

作者引用：按用户授权免除；未知值没有记为0。

绘图工具：unknown。未发现针对所选M/F图的具体绘图工具披露；一般写作/代码AI声明不作绘图工具推断。

裁剪坐标：[107, 80, 506, 209]，PDF点、左上角原点。

Figure 1: Motivation of UCMIT. Unlike position-wise fusion under cross-modal misalignment,
UCMIT retrieves complementary thermal evidence from plausible nearby locations and reconstructs
it in RGB coordinates before representation update.

最早核实视觉日期：2026-10-03；arXiv v1：未匹配；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

完整arXiv标题检索未找到匹配；补充标题检索未找到可确认的更早同稿。证据为公开投稿时间代理。

公开日期 · 论文入口

作者引用：按用户授权免除；未知值没有记为0。

绘图工具：unknown。未发现针对所选M/F图的具体绘图工具披露；一般写作/代码AI声明不作绘图工具推断。

裁剪坐标：[106, 79, 507, 242]，PDF点、左上角原点。

Figure 2: Overview of SEER. SEER converts regional VLM evidence into executable OVSS label
propagation. From a fixed prediction, it selects ambiguous proposals, derives score-based candidates,
and verifies them with multi-view closed-set reasoning. Score support and vocabulary legality certify
the verified label, and the OVSS model controls its pixel propagation.

最早核实视觉日期：2026-10-03；arXiv v1：未匹配；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

完整arXiv标题检索未找到匹配；补充标题检索未找到可确认的更早同稿。证据为公开投稿时间代理。

公开日期 · 论文入口

作者引用：按用户授权免除；未知值没有记为0。

绘图工具：unknown。未发现针对所选M/F图的具体绘图工具披露；一般写作/代码AI声明不作绘图工具推断。

裁剪坐标：[114, 91, 343, 208]，PDF点、左上角原点。

Figure 5: Residual failure cases. Mixed proposals, thin struc-
tures, and weak score support can leave errors near boundaries.
Additional examples appear in Appendix K.

最早核实视觉日期：2026-10-03；arXiv v1：未匹配；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

完整arXiv标题检索未找到匹配；补充标题检索未找到可确认的更早同稿。证据为公开投稿时间代理。

公开日期 · 论文入口

作者引用：按用户授权免除；未知值没有记为0。

绘图工具：unknown。未发现针对所选M/F图的具体绘图工具披露；一般写作/代码AI声明不作绘图工具推断。

裁剪坐标：[122, 79, 490, 283]，PDF点、左上角原点。

Figure 2: Overview of SHAPE. (a) The Edge Expert uses relation-level VFM support to guide
CLIP feature aggregation. (b) The Node Expert uses a patch-specific VFM neighborhood to refine
each score. (c) The Region Expert uses an agreement-gated regional support to reduce residual
fragmentation. Together, the three experts organize VFM structure into complementary supports for
feature aggregation, patch-specific correction, and regional consistency.

最早核实视觉日期：2026-10-03；arXiv v1：未匹配；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

完整arXiv标题检索未找到匹配；补充标题检索未找到可确认的更早同稿。证据为公开投稿时间代理。

公开日期 · 论文入口

作者引用：按用户授权免除；未知值没有记为0。

绘图工具：unknown。未发现针对所选M/F图的具体绘图工具披露；一般写作/代码AI声明不作绘图工具推断。

裁剪坐标：[106, 79, 507, 189.5]，PDF点、左上角原点。

Figure 1: Patch noise and region noise. Patch noise refers to isolated predictions that disagree with
surrounding patches, as seen in the cows (bottom), whereas region noise occurs when a semantic
region is split into inconsistent labels, as shown in the cow (top) and cats (middle). SHAPE reduces
both types of noise by selectively exploiting structural support through its Edge, Node, and Region
Experts at relation, patch, and region levels, respectively.

最早核实视觉日期：2026-10-03；arXiv v1：未匹配；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

完整arXiv标题检索未找到匹配；补充标题检索未找到可确认的更早同稿。证据为公开投稿时间代理。

公开日期 · 论文入口

作者引用：按用户授权免除；未知值没有记为0。

绘图工具：generative_AI_figure_editing_unspecified_tool。作者声明用生成式AI编辑科学图；工具名及逐图对应未披露。

裁剪坐标：[106, 79, 507, 299]，PDF点、左上角原点。

Figure 2: Overview of CVRoute. (a) Street-view semantics are routed to satellite-defined regions
for satellite-space prediction. (b) Cross-view entities are matched over multiple pose hypotheses and
fused through pose-marginalized nullable routing. (c) A training-only privileged relationship teacher
guides early routing and is gradually annealed out; inference uses only predicted relationships.

最早核实视觉日期：2026-10-03；arXiv v1：未匹配；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

完整arXiv标题检索未找到匹配；补充标题检索未找到可确认的更早同稿。证据为公开投稿时间代理。

公开日期 · 论文入口

作者引用：按用户授权免除；未知值没有记为0。

绘图工具：generative_AI_figure_editing_unspecified_tool。作者声明用生成式AI编辑科学图；工具名及逐图对应未披露。

裁剪坐标：[106, 79, 507, 229]，PDF点、左上角原点。

Figure 1: Motivation for semantic routing under pose uncertainty. Pose errors weaken the spatial
correspondence used by alignment-based cross-view fusion, motivating CVRoute to associate street-
view semantic evidence with satellite regions through learned cross-view relationships.

最早核实视觉日期：2026-10-03；arXiv v1：2026-10-13；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

固定OpenReview投稿PDF于10月3日公开；匹配的arXiv v1较后，不用于提前该视觉版本日期。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Konrad Schindler；47613；快照2026-10-21。

绘图工具：unknown。未发现针对所选M/F图的具体绘图工具披露；一般写作/代码AI声明不作绘图工具推断。

裁剪坐标：[111, 82, 501, 273]，PDF点、左上角原点。

Figure 2: Training and inference of S3-PnP. Purple arrows indicate full-space learning with P = 0,
s
P = I, and β = 0. Orange arrows indicate structure-guided learning: D is trained and frozen, and
g s s
its correction is enabled to collect a new state training set for D . At inference, active corrections are
g
evaluated at the same ADMM input and combined before a single analytical prior update. Snowflakes
denote frozen networks.

最早核实视觉日期：2026-10-03；arXiv v1：2026-10-13；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

固定OpenReview投稿PDF于10月3日公开；匹配的arXiv v1较后，不用于提前该视觉版本日期。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Konrad Schindler；47613；快照2026-10-21。

绘图工具：unknown。未发现针对所选M/F图的具体绘图工具披露；一般写作/代码AI声明不作绘图工具推断。

裁剪坐标：[121, 81, 502, 181]，PDF点、左上角原点。

Figure 1: Two sources of mismatch when using pretrained image priors in scientific imaging:
differences between the pretraining and target image domains, and differences between score-model
training inputs and ADMM denoiser inputs.

最早核实视觉日期：2026-10-03；arXiv v1：2026-05-17；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

当前整体布局压缩调整，不把旧构图日期赋给投稿图

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Qibin Hou；34230；快照2026-10-21。

绘图工具：unknown。未发现针对所选M/F图的具体绘图工具披露；一般写作/代码AI声明不作绘图工具推断。

裁剪坐标：[106, 79, 507, 271]，PDF点、左上角原点。

Figure 4: (a) The overall framework of DFC-IR. (b) FCMS for adaptive DFC tokenization and
spectral token extraction. (c) MMD for token-guided feature modulation and aggregation.

最早核实视觉日期：2026-10-03；arXiv v1：2026-05-17；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

v1六组变四组，图标与构图重绘

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Qibin Hou；34230；快照2026-10-21。

绘图工具：unknown。未发现针对所选M/F图的具体绘图工具披露；一般写作/代码AI声明不作绘图工具推断。

裁剪坐标：[106, 189, 507, 370]，PDF点、左上角原点。

Figure 1: Motivation of the proposed DFC-guided restoration paradigm. (a) Degraded images and residual frequency-domain energy maps; (b) DFCs derived from residual-to-degraded energy ratios; (c) Adaptive DFC tokenization into band-wise spectral tokens, where same-band tokens across degradations form reusable spectral priors; (d) Token-guided degradation-aware restoration.

最早核实视觉日期：2026-10-03；arXiv v1：未匹配；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

完整arXiv标题检索未找到匹配；补充标题检索未找到可确认的更早同稿。证据为公开投稿时间代理。

公开日期 · 论文入口

作者引用：按用户授权免除；未知值没有记为0。

绘图工具：unknown。未发现针对所选M/F图的具体绘图工具披露；一般写作/代码AI声明不作绘图工具推断。

裁剪坐标：[107, 79, 506, 266]，PDF点、左上角原点。

Figure 6: Overview of FAIR’s two-stage training. Stage 1 learns a general restoration prior. Stage 2
conditions restoration on region control R and guidance P = (t, v), using region-aware supervision
to bind the guidance to its target region.

最早核实视觉日期：2026-10-03；arXiv v1：未匹配；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

完整arXiv标题检索未找到匹配；补充标题检索未找到可确认的更早同稿。证据为公开投稿时间代理。

公开日期 · 论文入口

作者引用：按用户授权免除；未知值没有记为0。

绘图工具：unknown。未发现针对所选M/F图的具体绘图工具披露；一般写作/代码AI声明不作绘图工具推断。

裁剪坐标：[107, 79, 506, 266]，PDF点、左上角原点。

Figure 1: The “last-mile” problem in generative image restoration. State-of-the-art methods (Yu
et al., 2024; Wu et al., 2024b; Lin et al., 2025) produce globally plausible reconstructions from
severely degraded inputs, yet localized failures persist. Despite their small spatial extent, these
failures dominate perceived quality and practical acceptability.

最早核实视觉日期：2026-05-29；arXiv v1：2026-05-29；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

v1匹配

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Lei Zhang；137335；快照2026-10-21。

绘图工具：unknown。生成式图像模型用于数据集目标图生成，属于研究方法；不等同于M/F示意图绘制工具。

裁剪坐标：[106, 80, 507, 289]，PDF点、左上角原点。

Figure 3: Overview of the GGT-100K construction pipeline. We collect diverse real-world LQ images,
evaluate MFMs for HQ target generation, and apply multi-stage quality control to build the dataset.

最早核实视觉日期：2026-05-29；arXiv v1：2026-05-29；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

v1匹配

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Lei Zhang；137335；快照2026-10-21。

绘图工具：unknown。生成式图像模型用于数据集目标图生成，属于研究方法；不等同于M/F示意图绘制工具。

裁剪坐标：[106, 80, 507, 298]，PDF点、左上角原点。

Figure 2: Qualitative comparison of FoundIR (Li et al., 2025) and Qwen-Image-Edit (Wu et al., 2025)
trained/finetuned on different datasets. FoundIR shows limited generalization to real rain and haze,
even after finetuned on existing restoration datasets. Qwen-Image-Edit produces sharper details with
its strong generative prior but may introduce hallucinations or color shifts, while finetuning on existing
datasets weakens its detail generation ability. GGT-100K significantly improves the generalization
capability of them to real-world degradations, enhancing visual details while preserving scene fidelity.

最早核实视觉日期：2026-10-03；arXiv v1：2026-10-07；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

固定OpenReview投稿PDF于10月3日公开；匹配的arXiv v1较后，不用于提前该视觉版本日期。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Lei Zhang；137335；快照2026-10-21。

绘图工具：unknown。生成式模型是复原执行器，属于研究方法；不能由此推断M/F框图制作工具。

裁剪坐标：[106, 79, 507, 335]，PDF点、左上角原点。

Figure 2: Comparison between (a) prior agentic IR pipelines (top) and (b) HarnessIR (bottom).
Prior methods sequentially apply task-specific restoration models planned by VLMs to produce the
output. HarnessIR instead uses auxiliary tools for image-specific evidence, composes a restoration
instruction for MFM, and invokes verification-driven re-execution when necessary.

最早核实视觉日期：2026-10-03；arXiv v1：2026-10-07；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

固定OpenReview投稿PDF于10月3日公开；匹配的arXiv v1较后，不用于提前该视觉版本日期。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Lei Zhang；137335；快照2026-10-21。

绘图工具：unknown。生成式模型是复原执行器，属于研究方法；不能由此推断M/F框图制作工具。

裁剪坐标：[106, 80, 507, 205]，PDF点、左上角原点。

Figure 3: Direct use of MFMs may yield higher NR-IQA scores than GT, even when the image
content is largely altered. NR-IQA should not be used as an independent metric for IR performance.

最早核实视觉日期：2026-08-31；arXiv v1：2026-08-31；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

v1冻结骨干、能力路由与专家组合的图形元素和连接匹配

历史PDF · 引用证据 · 公开日期 · 论文入口

作者引用：Lizhuang Ma；13785；快照2026-10-28。

绘图工具：unknown。作者声明生成式AI协助Fig.5；本库所选为Fig.1和Fig.2，未将Fig.5披露推及M/F。

裁剪坐标：[107, 80, 505, 241]，PDF点、左上角原点。

Figure 2: Overview of the proposed RestoreMore framework. A pretrained restoration model is
preserved as a frozen capability anchor, and lightweight residual expansion modules are inserted at
selected feature stages. Inside each expansion module, a degradation descriptor is first constructed
from spatial and frequency statistics. Top-M capability selection is then performed over accumu-
lated capability prototypes and routers. The selected routers further retrieve complementary experts
from the expert-key bank through query–key matching. Frozen historical components and trainable
current-task components are indicated by snowflake and flame icons, respectively.

最早核实视觉日期：2026-08-31；arXiv v1：2026-08-31；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

v1固定集合与持续扩展对照的布局、图片及配色匹配

历史PDF · 引用证据 · 公开日期 · 论文入口

作者引用：Lizhuang Ma；13785；快照2026-10-28。

绘图工具：unknown。作者声明生成式AI协助Fig.5；本库所选为Fig.1和Fig.2，未将Fig.5披露推及M/F。

裁剪坐标：[107, 80, 505, 264]，PDF点、左上角原点。

Figure 1: Comparison of existing restoration strategies and RestoreMore. Existing methods are
limited to a fixed degradation set and typically require joint retraining with historical and new data
when new degradations arrive. In contrast, RestoreMore continually expands a pretrained restorer
by adding new experts using only newly arriving task data.

最早核实视觉日期：2026-10-03；arXiv v1：2026-10-14；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

固定OpenReview投稿PDF于10月3日公开；匹配的arXiv v1较后，不用于提前该视觉版本日期。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Lefei Zhang；18509；快照2026-10-21。

绘图工具：unknown。未发现针对所选M/F图的具体绘图工具披露；一般写作/代码AI声明不作绘图工具推断。

裁剪坐标：[106, 80, 507, 298]，PDF点、左上角原点。

Figure 2: Overview of PreActIR. The world model predicts action consequences before execution,
while the predict–act–verify controller executes and verifies the selected action with commit or
rollback. A tool-evolution loop performs candidate-frontier audit and targeted tool adaptation.

最早核实视觉日期：2026-10-03；arXiv v1：2026-10-14；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

固定OpenReview投稿PDF于10月3日公开；匹配的arXiv v1较后，不用于提前该视觉版本日期。

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Lefei Zhang；18509；快照2026-10-21。

绘图工具：unknown。未发现针对所选M/F图的具体绘图工具披露；一般写作/代码AI声明不作绘图工具推断。

裁剪坐标：[106, 81, 507, 270]，PDF点、左上角原点。

Figure 1: Motivation of PreActIR. Existing agentic restoration is largely reactive: tool effects are
observed only after execution, while decisions remain restricted to the available action space. Pre-
ActIR predicts action consequences before execution.

最早核实视觉日期：2026-10-03；arXiv v1：2026-09-25；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

与9月v1相比模型堆叠、标注、连接表现有调整，保守使用投稿公开日期

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Zheng-Jun Zha；30479；快照2026-10-21。

绘图工具：unknown。未发现针对所选M/F图的具体绘图工具披露；一般写作/代码AI声明不作绘图工具推断。

裁剪坐标：[106, 79, 507, 241]，PDF点、左上角原点。

Figure 4: Overview of PhoenixSR. HDA adapts real-score estimation, tracks the evolving student
distribution, and anchors paired fidelity. DRW reweights sample-wise guidance according to residual-
direction consistency.

最早核实视觉日期：2026-09-25；arXiv v1：2026-09-25；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

v1特征热图及局部失配示例匹配

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Zheng-Jun Zha；30479；快照2026-10-21。

绘图工具：unknown。未发现针对所选M/F图的具体绘图工具披露；一般写作/代码AI声明不作绘图工具推断。

裁剪坐标：[312, 460, 507, 564]，PDF点、左上角原点。

Figure 2: Motivation. CKA (Kornblith et al.,
2019) (left) reveals weak feature correspondence
between DBSR and DFSR, while residual and
zoom-in comparisons (right) show local discrep-
ancies between DBSR outputs and paired HR
targets.

最早核实视觉日期：2026-09-29；arXiv v1：2026-09-29；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

v1匹配

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Guangming Lu；13904；快照2026-10-21。

绘图工具：unknown; authors_deny_generative_AI_figures。作者明确否认使用生成式AI制作图；实际绘图软件未披露。

裁剪坐标：[106, 45, 507, 281]，PDF点、左上角原点。

Figure 2: Overview of RwF. Filter localization (left): parameter-space integrated gradients rank
layers, then filters within them, and the selected set is expanded with high-IG, low-similarity filters.
Filter learning (right): the localized filters are regenerated from a filter bank by factorized matrices
with cross-task attention and assembled back into the frozen backbone.

最早核实视觉日期：2026-09-29；arXiv v1：2026-09-29；ICLR 2027 frozen anonymous submission PDF; retrieved 2026-10-28 UTC

v1匹配

历史PDF · 下载引用证据快照 · 公开日期 · 论文入口

作者引用：Guangming Lu；13904；快照2026-10-21。

绘图工具：unknown; authors_deny_generative_AI_figures。作者明确否认使用生成式AI制作图；实际绘图软件未披露。

裁剪坐标：[106, 60, 507, 152]，PDF点、左上角原点。

Figure 1: (a) LoRA updates every filter of each adapted layer through low-rank matrices. (b) RwF
localizes the few degradation-critical filters of each task and regenerates only them from a shared
filter bank, leaving the rest of the backbone frozen.