Research map / Detection and segmentation / Weakly supervised detection and segmentation
Adapting Object Size Variance and Class Imbalance for Semi-supervised 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 the proposed adaptive pseudo labeling and multi-scale consistency method instead of the strongest listed baseline | Semi-supervised Faster R-CNN · MS-COCO val2017, 5% labeled | mAP: 32.50±0.08 -> 34.58±0.23 | Helped |
| use the proposed adaptive pseudo labeling and multi-scale consistency method instead of the strongest listed baseline | Semi-supervised Faster R-CNN · MS-COCO val2017, 10% labeled | mAP: 36.22±0.18 -> 37.50±0.14 | Helped |
| use the proposed adaptive pseudo labeling and multi-scale consistency method instead of the strongest listed baseline | Semi-supervised Faster R-CNN · MS-COCO val2017, 1% labeled | mAP: 23.02±0.23 -> 24.04±0.69 | Helped |
| use the proposed adaptive pseudo labeling and multi-scale consistency method instead of the strongest listed baseline | Semi-supervised Faster R-CNN · MS-COCO val2017, 100% labeled | mAP: 46.10 -> 46.20 | Helped |
| use the proposed method instead of the strongest listed baseline | Semi-supervised Faster R-CNN · PASCAL VOC07 test | mAP: 81.72 -> 84.70 | Helped |
| add feature-level consistency learning (ablation) | Faster R-CNN on MS-COCO with 10% labeled data · MS-COCO, 10% labeled | mAP: 27.2 -> 35.1 | Helped |
| add prediction-level consistency learning to feature-level consistency learning (ablation) | Faster R-CNN on MS-COCO with 10% labeled data · MS-COCO, 10% labeled | mAP: 35.1 -> 35.8 | Helped |
| add adaptive pseudo labeling (ablation) | Faster R-CNN with feature- and prediction-level consistency · MS-COCO, 10% labeled | mAP: 35.8 -> 36.9 | Helped |
| add probability sharpening (ablation) | Faster R-CNN with consistency learning and APL · MS-COCO, 10% labeled | mAP: 36.9 -> 37.2 | Helped |
| vary the TopK class count in adaptive pseudo labeling; the paper reports K = 3 as the best choice (ablation) | Adaptive pseudo labeling · MS-COCO, 10% labeled | mAP | Mixed |
| vary the feature-level consistency loss weight β; the paper reports β = 2.0 as best (ablation) | Feature-level consistency learning · MS-COCO, 10% labeled | mAP | Mixed |
Closest papers by meaning
- Rethinking Pseudo Labels for Semi-supervised Object Detection (AAAI 2022 · similarity 0.96)
- Mind the Gap: Polishing Pseudo Labels for Accurate Semi-supervised Object Detection (AAAI 2023 · similarity 0.92)
- Consistent-Teacher: Towards Reducing Inconsistent Pseudo-Targets in Semi-Supervised Object Detection (CVPR 2023 · similarity 0.92)
- Semi-supervised Object Detection with Adaptive Class-Rebalancing Self-Training (AAAI 2022 · similarity 0.92)
- Instant-Teaching: An End-to-End Semi-Supervised Object Detection Framework (CVPR 2021 · similarity 0.91)
- Multi-clue Consistency Learning to Bridge Gaps Between General and Oriented Object in Semi-supervised Detection (AAAI 2025 · 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.