Research map / Detection and segmentation / Weakly supervised detection and segmentation
Online Refinement of Low-Level Feature Based Activation Map for Weakly Supervised Object Localization
Jinheng Xie, Cheng Luo, Xiangping Zhu, Ziqi Jin, Weizeng Lu, Linlin Shen
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 the evaluator supervised by evaluation loss in stage two (ablation) | stage-one activation map generator · CUB-200-2011 | CorLoc: 64.52 -> 71.35 | Improved |
| add weighted entropy loss to evaluation loss (ablation) | ORNet stage two · CUB-200-2011 | CorLoc: 71.35 -> 81.68 | Improved |
| add weighted entropy loss to evaluation loss (ablation) | ORNet stage two · CUB-200-2011 | Top-1 LocErr: 44.36 -> 36.05 | Improved |
| add area loss to evaluation loss alone (ablation) | ORNet stage two · CUB-200-2011 | CorLoc: 71.35 -> 72.11 | No clear effect |
| add attentive erasing to evaluation loss alone (ablation) | ORNet stage two · CUB-200-2011 | CorLoc: 71.35 -> 69.14 | Hurt |
| add area loss and attentive erasing together (ablation) | Le · CUB-200-2011 | CorLoc: 71.35 -> 76.68 | Improved |
| add weighted entropy loss on top of area loss and attentive erasing (ablation) | Le+La+AE · CUB-200-2011 | CorLoc: 76.68 -> 86.19 | Improved |
| add area loss on top of weighted entropy loss and attentive erasing (ablation) | Le+Lw+AE · CUB-200-2011 | CorLoc: 85.09 -> 86.19 | Improved |
| add attentive erasing on top of weighted entropy loss and area loss (ablation) | Le+Lw+La · CUB-200-2011 | CorLoc: 80.10 -> 86.19 | Improved |
| use weighted entropy loss instead of plain mean entropy loss (ablation) | Le+La+AE · CUB-200-2011 | CorLoc: 72.89 -> 86.19 | Improved |
| use weighted entropy loss instead of plain mean entropy loss (ablation) | Le+La+AE · CUB-200-2011 | Top-1 LocErr: 42.53 -> 32.26 | Improved |
| vary threshold from 0.2 to 0.8 (ablation) | ORNet · CUB-200-2011 and ImageNet-1K | CorLoc | No clear effect |
Closest papers by meaning
- Rethinking Class Activation Mapping for Weakly Supervised Object Localization (ECCV 2020 · similarity 0.95)
- Background-Aware Classification Activation Map for Weakly Supervised Object Localization (TPAMI 2023 · similarity 0.93)
- Bridging the Gap Between Classification and Localization for Weakly Supervised Object Localization (CVPR 2022 · similarity 0.93)
- CREAM: Weakly Supervised Object Localization via Class RE-Activation Mapping (CVPR 2022 · similarity 0.92)
- Evaluating Weakly Supervised Object Localization Methods Right (CVPR 2020 · similarity 0.92)
- Evaluation for Weakly Supervised Object Localization: Protocol, Metrics, and Datasets (TPAMI 2022 · similarity 0.92)
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.
