Research map / Detection and segmentation / Weakly supervised detection and segmentation
W2N: Switching from Weak Supervision to Noisy Supervision for Object Detection
Zitong Huang, Yiping Bao, Bowen Dong, Erjin Zhou, Wangmeng Zuo
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 W2N framework | LBBA · PASCAL VOC 2007 test | mAP: 59.1 -> 68.6 | Improved |
| add W2N framework | CASD · PASCAL VOC 2007 test | mAP: 54.0 -> 65.4 | Improved |
| add W2N framework | OICR+REG · PASCAL VOC 2007 test | mAP: 48.3 -> 57.0 | Improved |
| add W2N framework | CASD · PASCAL VOC 2007 trainval | CorLoc: 68.5 -> 80.1 | Improved |
| add W2N framework | LBBA · PASCAL VOC 2007 trainval | CorLoc: 72.7 -> 83.5 | Improved |
| retrain Faster R-CNN on pseudo ground-truth only (ablation) | LBBA · PASCAL VOC 2007 test | mAP: 59.1 -> 59.4 | No clear effect |
| add localization adaptation module (ablation) | LBBA · PASCAL VOC 2007 test | mAP: 59.1 -> 60.3 | Improved |
| add semi-supervised learning module (ablation) | LBBA · PASCAL VOC 2007 test | mAP: 59.1 -> 66.1 | Improved |
| combine localization adaptation with semi-supervised learning (ablation) | LBBA · PASCAL VOC 2007 test | mAP: 66.1 -> 67.0 | Improved |
| add iterative training (T=2) (ablation) | LBBA+W2N · PASCAL VOC 2007 test | mAP: 67 -> 68.6 | Improved |
| add iterative training (T=1) (ablation) | OICR+REG+W2N · PASCAL VOC 2007 test | mAP: 56.8 -> 57.0 | No clear effect |
| use instance-level split instead of image-level split (ablation) | LBBA+W2N · PASCAL VOC 2007 test | mAP: 65.4 -> 66.8 | Improved |
Closest papers by meaning
- W2F: A Weakly-Supervised to Fully-Supervised Framework for Object Detection (CVPR 2018 · similarity 0.93)
- UWSOD: Toward Fully-Supervised-Level Capacity Weakly Supervised Object Detection (NeurIPS 2020 · similarity 0.92)
- Object-Aware Instance Labeling for Weakly Supervised Object Detection (ICCV 2019 · similarity 0.92)
- Salvage of Supervision in Weakly Supervised Object Detection (CVPR 2022 · similarity 0.92)
- GradingNet: Towards Providing Reliable Supervisions for Weakly Supervised Object Detection by Grading the Box Candidates (AAAI 2021 · similarity 0.91)
- Weakly Supervised Few-Shot Object Detection with DETR (AAAI 2024 · similarity 0.90)
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.
