Research map / Detection and segmentation / Weakly supervised detection and segmentation
Co-mining: Self-Supervised Learning for Sparsely Annotated Object Detection
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| use Co-mining instead of BRL | RetinaNet, ResNet-101 · COCO-50miss, COCO-2017 val | AP: 32.7 -> 33.9 | Helped |
| use Co-mining instead of Part-aware | RetinaNet, ResNet-101 · COCO-50miss, COCO-2017 val | AP: 28.4 -> 33.9 | Helped |
| add Co-mining rather than train on original inputs only | RetinaNet · COCO-2017 val, Extreme | AP: 20.9 -> 23.0 | Helped |
| add pseudo-label co-generation (ablation) | RetinaNet · COCO-2017 val, Hard | AP: 30.4 -> 30.9 | Helped |
| add a Siamese network with original and augmented inputs to co-generation (ablation) | RetinaNet · COCO-2017 val, Hard | AP: 30.9 -> 31.8 | Helped |
| use score threshold 0.4 for pseudo-labels (ablation) | RetinaNet Co-mining · COCO-2017 val, Hard | AP: Nan | Hurt |
| use score threshold 0.5 instead of 0.6 (ablation) | RetinaNet Co-mining · COCO-2017 val, Hard | AP: 31.8 -> 30.6 | Hurt |
| raise score threshold to 0.7 or 0.8 (ablation) | RetinaNet Co-mining · COCO-2017 val, Hard | AP: 31.8 -> 30.7 | Hurt |
| use color augmentation instead of no augmentation (ablation) | RetinaNet Co-mining · COCO-2017 val, Hard | AP: 30.6 -> 31.8 | Helped |
| use color augmentation rather than blur augmentation (ablation) | RetinaNet Co-mining · COCO-2017 val, Hard | AP: 31.2 -> 31.8 | Helped |
Closest papers by meaning
- SIOD: Single Instance Annotated per Category per Image for Object Detection (CVPR 2022 · similarity 0.88)
- Towards Human-Machine Cooperation: Self-Supervised Sample Mining for Object Detection (CVPR 2018 · similarity 0.87)
- As Pseudo-Label Free as Possible: Leveraging Adaptive Feature Generation for Sparsely Annotated Object Detection (AAAI 2025 · similarity 0.87)
- Object Instance Mining for Weakly Supervised Object Detection (AAAI 2020 · similarity 0.87)
- S²Teacher: Step-by-step Teacher for Sparsely Annotated Oriented Object Detection (AAAI 2026 · similarity 0.87)
- Instance-Aware, Context-Focused, and Memory-Efficient Weakly Supervised Object Detection (CVPR 2020 · 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.