AI Paper Atlas

Research map / Detection and segmentation / Weakly supervised detection and segmentation

Class Token as Proxy: Optimal Transport-assisted Proxy Learning for Weakly Supervised Semantic Segmentation

Jian Wang, Tianhong Dai, Bingfeng Zhang, Siyue Yu, Eng Gee Lim, Jimin Xiao

ICCV 2025 · Detection and segmentation › Weakly supervised detection and segmentation · approach: CAM refinement and prototypes · 3 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add OT-assisted proxy learning (ablation)MCTformer+ · PASCAL VOC 2012 trainmIoU: 68.8 -> 71.5Improved
add OT-assisted proxy-patch contrastive learning (Lppc) directly on class tokens (ablation)MCTformer+ · PASCAL VOC 2012 trainmIoU: 68.8 -> 70.7Improved
add Lppc on top of proxy learning (ablation)OTPL · PASCAL VOC 2012 trainmIoU: 71.5 -> 73.4Improved
add CAM-based marginal constraint in OT (ablation)OTPL proxy learning · PASCAL VOC 2012 trainmIoU: 71.0 -> 71.5Improved
select positive pairs using both CAM and OT plan instead of CAM only (ablation)Lppc · PASCAL VOC 2012 trainmIoU: 72.5 -> 73.4Improved
select positive pairs using both CAM and OT plan instead of OT only (ablation)Lppc · PASCAL VOC 2012 trainmIoU: 72.7 -> 73.4Improved
integrate OTPL into CLIP-ESCLIP-ES · PASCAL VOC 2012 trainCAM mIoU: 70.8 -> 74.8Improved
integrate OTPL into CLIP-ESCLIP-ES · PASCAL VOC 2012 trainpseudo-mask mIoU: 75.0 -> 79.2Improved
integrate OTPL into MCTformer+MCTformer+ · PASCAL VOC 2012 trainCAM mIoU: 68.8 -> 73.4Improved
integrate OTPL into CLIP-ESCLIP-ES · MS COCO 2014 valmIoU: 45.4 -> 47.6Improved
integrate OTPL into MCTformer+MCTformer+ · PASCAL VOC 2012 val (segmentation)mIoU: 74.0 -> 75.6Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy