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.