Research map / Detection and segmentation / Weakly supervised detection and segmentation
Category-aware Allocation Transformer for Weakly Supervised Object Localization
Zhiwei Chen, Jinren Ding, Liujuan Cao, Yunhang Shen, Shengchuan Zhang, Guannan Jiang, Rongrong Ji
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add Category-aware Stimulation Module (CSM) to baseline (ablation) | CATR (TS-CAM baseline) · CUB-200-2011 test | Top-1 Loc: 74.16 -> 77.59 | Improved |
| add Object Constraint Module (OCM) on top of CSM (ablation) | CATR · CUB-200-2011 test | Top-1 Loc: 77.59 -> 79.62 | Improved |
| set CSM channel width G to 8 (ablation) | CATR · CUB-200-2011 test | Top-1 Loc: 72.20 -> 77.59 (G=2 vs 8) | Improved |
| use small CSM width G of 2 or 4 instead of no CSM (ablation) | CATR · CUB-200-2011 test | Top-1 Loc: 74.16 -> 72.20 | Hurt |
| increase CSM width G to 32 (ablation) | CATR · CUB-200-2011 test | Top-1 Loc: 77.59 -> 74.18 | Hurt |
| aggregate top-K=9 transformer blocks in OCM (ablation) | CATR · CUB-200-2011 test | Top-1 Loc: 78.20 -> 79.62 | Improved |
| increase K beyond 9 up to 12 (ablation) | CATR · CUB-200-2011 test | Top-1 Loc: 79.62 -> 65.55 | Hurt |
| raise noise suppression percentile p beyond 10% (ablation) | CATR · CUB-200-2011 test | Top-1 Loc | Hurt |
| add automatic weighted loss mechanism (ablation) | CATR · CUB-200-2011 test | Top-1 Loc: 79.41 -> 79.62 | Improved |
| add automatic weighted loss mechanism (ablation) | CATR · ILSVRC val | Top-1 Loc: 56.13 -> 56.90 | Improved |
| replace TS-CAM with CATR on ILSVRC | TS-CAM · ILSVRC val | Top-1 Loc: 53.40 -> 56.90 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- LCTR — LCTR: On Awakening the Local Continuity of Transformer for Weakly Supervised Object Localization (AAAI 2022) · better on 2 of 2 shared results
- TS-CAM — TS-CAM: Token Semantic Coupled Attention Map for Weakly Supervised Object Localization (ICCV 2021) · better on 2 of 2 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
- CLIP-Driven Transformer for Weakly Supervised Object Localization (TPAMI 2025 · similarity 0.94)
- Spatial-Aware Token for Weakly Supervised Object Localization (ICCV 2023 · similarity 0.93)
- Weakly Supervised Object Localization via Transformer with Implicit Spatial Calibration (ECCV 2022 · similarity 0.92)
- LCTR: On Awakening the Local Continuity of Transformer for Weakly Supervised Object Localization (AAAI 2022 · similarity 0.91)
- Task-Aware Weakly Supervised Object Localization With Transformer (TPAMI 2022 · similarity 0.91)
- Cross-token Guidance Transformer for Weakly Supervised Object Localization (ECCV 2026 · similarity 0.91)
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.
