Research map / Detection and segmentation / Weakly supervised detection and segmentation
Improving Object Detection with Selective Self-Supervised Self-Training
Yandong Li, Di Huang, Danfeng Qin, Liqiang Wang, Boqing Gong
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 selective net to vanilla self-training (S2OD vs SOD) (ablation) | SOD · COCO-backpack | AP@[.5,.95]: 17.62 -> 18.28 | Improved |
| add the selective net to CSD (ablation) | CSD · COCO-backpack | AP@[.5,.95]: 17.87 -> 18.32 | Improved |
| add the selective net to CSD (ablation) | CSD · COCO-chair | AP@[.5,.95]: 30.19 -> 30.95 | Improved |
| add self-supervised loss on selected boxes (S4OD vs S2OD) (ablation) | S2OD · COCO-backpack | AP@[.5,.95]: 18.28 -> 19.48 | Improved |
| add self-supervised loss on selected boxes (S4OD vs S2OD) (ablation) | S2OD · COCO-chair | AP@[.5,.95]: 30.95 -> 31.74 | Improved |
| use image-to-image search instead of text-to-image search for web images (ablation) | S4OD · COCO-backpack | AP@[.5,.95]: 17.37 -> 19.48 | Improved |
| increase web images from 1/3 to full (ablation) | S4OD · COCO-backpack | AP@[.5,.95]: 18.40 -> 19.08 -> 19.48 | Improved |
| add a second self-training iteration (ablation) | S4OD · COCO-backpack | AP@[.5,.95]: 19.48 -> 19.52 | No clear effect |
| add vanilla self-training with web images (SOD vs BD) | BD · COCO-backpack | AP@[.5,.95]: 17.34 -> 17.62 | Mixed |
| train binary detector instead of 80-class detector (BD vs R101-FPN) | R101-FPN · COCO-backpack | AP@[.5,.95]: 16.45 -> 17.34 | Improved |
| add S4OD on relabeled data | BD · relabeled COCO-backpack | AP@[.5,.95]: 18.75 -> 20.27 | Improved |
| add selective net and self-supervised loss on VOC semi-supervised (S4OD vs CSD) | CSD · PASCAL VOC2007 test | AP: 74.7 -> 76.4 | Improved |
Closest papers by meaning
- A Survey of Self-Supervised and Few-Shot Object Detection (TPAMI 2022 · similarity 0.90)
- Exploring Bottom-Up and Top-Down Cues With Attentive Learning for Webly Supervised Object Detection (CVPR 2020 · similarity 0.89)
- Noise-Aware Fully Webly Supervised Object Detection (CVPR 2020 · similarity 0.88)
- Object Discovery and Representation Networks (ECCV 2022 · similarity 0.88)
- Towards Human-Machine Cooperation: Self-Supervised Sample Mining for Object Detection (CVPR 2018 · similarity 0.88)
- Search and Detect: Training-Free Long Tail Object Detection via Web-Image Retrieval (CVPR 2025 · similarity 0.86)
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.
