Research map / Detection and segmentation / Weakly supervised detection and segmentation
Weakly Supervised Object Localization Using Size Estimates
Miaojing Shi, Vittorio Ferrari
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 size-ordered curriculum to baseline MIL (ablation) | Baseline MIL · PASCAL VOC 2007 | CorLoc: 39.1 -> 46.3 | Improved |
| add size-ordered curriculum to baseline MIL (ablation) | Baseline MIL · PASCAL VOC 2007 test | mAP: 20.1 -> 24.9 | Improved |
| add size weighting of proposals on top of size order (ablation) | MIL + size order · PASCAL VOC 2007 | CorLoc: 46.3 -> 55.8 | Improved |
| add size weighting of proposals on top of size order (ablation) | MIL + size order · PASCAL VOC 2007 test | mAP: 24.9 -> 28.0 | Improved |
| add Fast R-CNN deep re-training iterations to size order and weighting (ablation) | MIL + size order + weighting · PASCAL VOC 2007 | CorLoc: 55.8 -> 60.9 | Improved |
| add Fast R-CNN deep re-training iterations to size order and weighting (ablation) | MIL + size order + weighting · PASCAL VOC 2007 test | mAP: 28.0 -> 36.0 | Improved |
| add deep Fast R-CNN iterations to the baseline (ablation) | Baseline MIL · PASCAL VOC 2007 | CorLoc and mAP: +4.1 CorLoc, +4.6 mAP | Improved |
| use VGG16 instead of AlexNet in deep Fast R-CNN re-training | Full method · PASCAL VOC 2007 | CorLoc / mAP: 60.9 -> 64.7 CorLoc, 36.0 -> 37.2 mAP | Improved |
| use class-generic size estimator instead of class-specific (ablation) | MIL + size order + weighting · PASCAL VOC 2007 | CorLoc: 55.8 -> 48.4 | Hurt |
| use across-class size estimator instead of class-specific (ablation) | MIL + size order + weighting · PASCAL VOC 2007 | CorLoc: 55.8 -> 45.8 | Hurt |
| use across-class size estimator for size order only (ablation) | MIL + size order · PASCAL VOC 2007 | CorLoc: 46.3 -> 45.0 | No clear effect |
| use class-generic estimator with deep re-training (ablation) | Full method · PASCAL VOC 2007 test | mAP: 36.0 -> 32.2 | Hurt |
Closest papers by meaning
- Pairwise Similarity Knowledge Transfer for Weakly Supervised Object Localization (ECCV 2020 · similarity 0.93)
- Evaluation for Weakly Supervised Object Localization: Protocol, Metrics, and Datasets (TPAMI 2022 · similarity 0.93)
- Evaluating Weakly Supervised Object Localization Methods Right (CVPR 2020 · similarity 0.93)
- Rethinking the Route Towards Weakly Supervised Object Localization (CVPR 2020 · similarity 0.92)
- Weakly Supervised Object Localization As Domain Adaption (CVPR 2022 · similarity 0.91)
- Boosting Weakly Supervised Object Localization and Segmentation With Domain Adaption (TPAMI 2024 · 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.
