Research map / Detection and segmentation / Weakly supervised detection and segmentation
Bridging the Gap Between Classification and Localization for Weakly Supervised Object Localization
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| align feature directions and apply attentive dropout | CAM-based WSOL classifier · CUB-200-2011 test, VGG16 | Top-1 Loc: 46.95 -> 70.83 | Helped |
| add attentive dropout consistency (ablation) | feature direction alignment · CUB-200-2011 test, VGG16 | Top-1 Loc: 62.27 -> 70.83 | Helped |
| add feature direction alignment using similarity loss (ablation) | CAM-based WSOL classifier · CUB-200-2011 test, VGG16 | Top-1 Loc: 46.95 -> 56.66 | Helped |
| add norm loss to similarity loss (ablation) | feature direction alignment · CUB-200-2011 test, VGG16 | Top-1 Loc: 56.66 -> 62.27 | Helped |
| add attentive dropout to similarity loss (ablation) | feature direction alignment · CUB-200-2011 test, VGG16 | Top-1 Loc: 56.66 -> 63.00 | Helped |
| replace attentive dropout with EIL (ablation) | feature direction alignment · CUB-200-2011 test, VGG16 | Top-1 Loc: 70.83 -> 66.10 | Hurt |
| use the proposed method | WSOL classifier with VGG16 · ImageNet-1K validation | Top-1 Loc: 49.94 vs 49.56 | Mixed |
| use the proposed method | WSOL classifier with ResNet50 · ImageNet-1K validation | Top-1 Loc: 53.76 vs 54.83 | Hurt |
| set attentive-dropout constraint too strongly, with lambda_drop=4 (ablation) | proposed method · CUB-200-2011 test, VGG16 | localization performance | Hurt |
| use deterministic dropout with probability 1.0 instead of stochastic dropout (ablation) | attentive dropout · CUB-200-2011 test, VGG16 | localization performance | Hurt |
Closest papers by meaning
- Rethinking the Route Towards Weakly Supervised Object Localization (CVPR 2020 · similarity 0.92)
- Evaluating Weakly Supervised Object Localization Methods Right (CVPR 2020 · similarity 0.91)
- CREAM: Weakly Supervised Object Localization via Class RE-Activation Mapping (CVPR 2022 · similarity 0.90)
- Unveiling the Potential of Structure Preserving for Weakly Supervised Object Localization (CVPR 2021 · similarity 0.90)
- Shallow Feature Matters for Weakly Supervised Object Localization (CVPR 2021 · similarity 0.90)
- Coarse2Fine: Local Consistency Aware Re-prediction for Weakly Supervised Object Localization (AAAI 2023 · similarity 0.89)
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.