Research map / Detection and segmentation / Weakly supervised detection and segmentation
Zero-Annotation Object Detection with Web Knowledge Transfer
Qingyi Tao, Hao Yang, Jianfei Cai
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 DA stream alone to web-trained WSD (ablation) | WSD baseline VGG_M · VOC 2007 test | mAP: 21.5 -> 21.2 | No clear effect |
| add DA stream alone to web-trained WSD (ablation) | WSD baseline VGG16 · VOC 2007 test | mAP: 21.8 -> 21.0 | Hurt |
| add one self-training pseudo-label (ST) stream on top of DA | WSD+DA VGG_M · VOC 2007 test | mAP: 21.2 -> 24.2 | Improved |
| add one ST stream on top of DA | WSD+DA VGG16 · VOC 2007 test | mAP: 21.0 -> 25.0 | Improved |
| add second and third chained ST streams (ablation) | WSD+DA+ST VGG16 · VOC 2007 test | mAP: 25.0 -> 25.4 | Improved |
| add second and third chained ST streams (ablation) | WSD+DA+ST VGG_M · VOC 2007 test | mAP: 24.2 -> 24.6 | Improved |
| add DA stream with foreground attention to WSD+3ST (ablation) | WSD+3ST VGG_M · VOC 2007 test | mAP: 23.5 -> 24.6 | Improved |
| add foreground attention to DA stream (ablation) | WSD+DA+3ST VGG_M · VOC 2007 test | mAP: 23.3 -> 24.6 | Improved |
| add DA without foreground attention (ablation) | WSD+3ST VGG_M · VOC 2007 test | mAP: 23.5 -> 23.3 | No clear effect |
| train ST stream simultaneously with WSD instead of isolated pseudo-label training (ablation) | WSD+DA+3ST VGG_M · VOC 2007 test | mAP: 22.5 -> 24.6 | Improved |
| apply full method (DA+3ST) to web-only baseline | WSD VGG_M · VOC 2012 test | mAP: 17.9 -> 21.7 | Improved |
| apply full method (DA+3ST) to web-only baseline | WSD VGG16 · VOC 2012 test | mAP: 19.2 -> 22.9 | Improved |
Closest papers by meaning
- Cross-Domain Weakly-Supervised Object Detection Through Progressive Domain Adaptation (CVPR 2018 · similarity 0.90)
- Boosting Weakly Supervised Object Detection with Progressive Knowledge Transfer (ECCV 2020 · similarity 0.90)
- Cross-Domain Adaptive Teacher for Object Detection (CVPR 2022 · similarity 0.89)
- Transferring Labels to Solve Annotation Mismatches Across Object Detection Datasets (ICLR 2024 · similarity 0.89)
- Self-Training and Adversarial Background Regularization for Unsupervised Domain Adaptive One-Stage Object Detection (ICCV 2019 · similarity 0.89)
- Large Self-Supervised Models Bridge the Gap in Domain Adaptive Object Detection (CVPR 2025 · similarity 0.87)
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.
