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.