Research map / Detection and segmentation / Weakly supervised detection and segmentation
Data-Uncertainty Guided Multi-Phase Learning for Semi-Supervised Object Detection
Zhenyu Wang, Yali Li, Ye Guo, Lu Fang, Shengjin Wang
Check 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 |
|---|---|---|---|
| Use data-uncertainty guided multi-phase training instead of the one-phase baseline | Faster RCNN · PASCAL VOC 2007 test | mAP: 75.6 -> 78.6 | Improved |
| Use data-uncertainty guided multi-phase training instead of the one-phase baseline | SSD300 · PASCAL VOC 2007 test | mAP: 71.8 -> 74.5 | Improved |
| Add the proposed method to DD | Faster RCNN, ResNet50 · COCO minival, co-35/co-80 | AP: 33.1 -> 35.2 | Improved |
| Add the proposed method to DD | Faster RCNN, ResNet50 · COCO minival, co-115/co-120 | AP: 37.9 -> 40.3 | Improved |
| Add two-phase learning (ablation) | Faster RCNN · PASCAL VOC 2007 test | mAP: 75.6 -> 76.1 | Improved |
| Add RoI Re-weighting to two-phase learning (ablation) | Faster RCNN · PASCAL VOC 2007 test | mAP: 76.1 -> 77.4 | Improved |
| Add model ensembling across phases (ablation) | Faster RCNN · PASCAL VOC 2007 test | mAP: 77.4 -> 78.6 | Improved |
| Add model ensembling across phases (ablation) | SSD300 · PASCAL VOC 2007 test | mAP: 72.3 -> 74.5 | Improved |
| Use overlap-and-similarity RoI re-weighting rather than the baseline (ablation) | Faster RCNN · PASCAL VOC 2007 test | mAP: 76.1 -> 77.4 | Improved |
| Use overlap-and-similarity RoI re-weighting rather than overlap-based Soft Sampling (ablation) | Faster RCNN · PASCAL VOC 2007 test | mAP: 76.6 -> 77.4 | Improved |
| Train phase 1 with easy unlabeled images (ablation) | Faster RCNN · PASCAL VOC 2007 test (easy) | mAP: 86.2 -> 86.7 | Improved |
| Add difficult unlabeled images in phase 2 (ablation) | Faster RCNN · PASCAL VOC 2007 test (easy) | mAP: 86.7 -> 86.4 | Hurt |
Closest papers by meaning
- Combating Noise: Semi-supervised Learning by Region Uncertainty Quantification (NeurIPS 2021 · similarity 0.91)
- Rethinking Pseudo Labels for Semi-supervised Object Detection (AAAI 2022 · similarity 0.89)
- Adapting Object Size Variance and Class Imbalance for Semi-supervised Object Detection (AAAI 2023 · similarity 0.88)
- Unbiased Teacher for Semi-Supervised Object Detection (ICLR 2021 · similarity 0.87)
- Multi-clue Consistency Learning to Bridge Gaps Between General and Oriented Object in Semi-supervised Detection (AAAI 2025 · similarity 0.87)
- RWMS: Reliable Weighted Multi-Phase for Semi-supervised Segmentation (AAAI 2024 · 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.
