Research map / Detection and segmentation / Weakly supervised detection and segmentation
Object-Aware Instance Labeling for Weakly Supervised Object Detection
Satoshi Kosugi, Toshihiko Yamasaki, Kiyoharu Aizawa
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 CAP labeling to OICR (ablation) | OICR · PASCAL VOC 2007 | mAP: 41.2 -> 45.6 | Improved |
| add CAP labeling to OICR (ablation) | OICR · PASCAL VOC 2007 | CorLoc: 60.6 -> 66.6 | Improved |
| add SRN labeling to OICR (ablation) | OICR · PASCAL VOC 2007 | mAP: 41.2 -> 45.1 | Improved |
| add SRN labeling to OICR (ablation) | OICR · PASCAL VOC 2007 | CorLoc: 60.6 -> 63.4 | Improved |
| add SRN labeling on top of CAP labeling (ablation) | CAP labeling · PASCAL VOC 2007 | mAP: 45.6 -> 47.6 | Improved |
| add CAP labeling on top of SRN labeling (ablation) | SRN labeling · PASCAL VOC 2007 | mAP: 45.1 -> 47.6 | Improved |
| add SRN labeling on top of CAP labeling (ablation) | CAP labeling · PASCAL VOC 2007 | CorLoc: 66.6 -> 66.7 | No clear effect |
| use context classification instead of simple mask-out (ablation) | Ours · PASCAL VOC 2007 | mAP: 47.1 -> 47.6 | Improved |
| use context classification instead of simple mask-out (ablation) | Ours · PASCAL VOC 2007 | CorLoc: 64.9 -> 66.7 | Improved |
| apply PGE pseudo ground-truth excavation to Ours + FRCNN | Ours + FRCNN · PASCAL VOC 2007 test (Fast R-CNN retrained) | mAP: 51.4 -> 52.1 on VOC 2007; 48.1 -> 47.9 on VOC 2012 | Mixed |
| Train Fast R-CNN on our localization instead of OICR with PGE | Fast R-CNN · PASCAL VOC 2012 test (Fast R-CNN retrained) | mAP: 47.3 -> 48.1 | Improved |
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.
- TS2C — TS2C: Tight Box Mining with Surrounding Segmentation Context for Weakly Supervised Object Detection (ECCV 2018) · better on 4 of 4 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
- Object Instance Mining for Weakly Supervised Object Detection (AAAI 2020 · similarity 0.93)
- W2N: Switching from Weak Supervision to Noisy Supervision for Object Detection (ECCV 2022 · similarity 0.92)
- Instance Mining with Class Feature Banks for Weakly Supervised Object Detection (AAAI 2021 · similarity 0.92)
- Object Discovery via Contrastive Learning for Weakly Supervised Object Detection (ECCV 2022 · similarity 0.92)
- Weakly Supervised Open-Vocabulary Object Detection (AAAI 2024 · similarity 0.90)
- Weakly Supervised Few-Shot Object Detection with DETR (AAAI 2024 · 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.
