Research map / Detection and segmentation / CNN features
Training Region-Based Object Detectors With Online Hard Example Mining
Check 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 |
|---|---|---|---|
| Replace heuristic RoI sampling with online selection of high-loss RoIs | Fast R-CNN, VGG16 · VOC07 test, trainval | mAP: 67.2 -> 69.9 | Helped |
| Replace heuristic RoI sampling with online selection of high-loss RoIs | Fast R-CNN, VGG16 · VOC12 test, trainval | mAP: 65.7 -> 69.8 | Helped |
| Replace heuristic RoI sampling with online selection of high-loss RoIs | Fast R-CNN, VGG16 · MS COCO test-dev 2015 | AP@[0.50:0.95]: 19.7 -> 22.6 | Helped |
| Add OHEM to multi-scale training/testing and iterative bounding-box regression | Fast R-CNN, VGG16 · VOC07 test, 07+12 trainval | mAP: 78.2 -> 78.9 | Helped |
| Add OHEM to multi-scale training/testing and iterative bounding-box regression | Fast R-CNN, VGG16 · VOC12 test, 07++12 trainval | mAP: 73.9 -> 76.3 | Helped |
| Add multi-scale training and testing (ablation) | Fast R-CNN, VGG16 · VOC07 test, trainval | mAP: 67.2 -> 68.6 | Helped |
| Add multi-scale testing (ablation) | Fast R-CNN, VGG16 · VOC07 test, trainval | mAP: 67.2 -> 68.4 | Helped |
| Add iterative bounding-box regression (ablation) | Fast R-CNN, VGG16 · VOC07 test, trainval | mAP: 67.2 -> 70.8 | Helped |
| Remove the bg_lo=0.1 hard-mining heuristic (ablation) | Fast R-CNN, VGGM · VOC07 test | mAP: 59.6 -> 57.2 | Hurt |
| Remove the bg_lo=0.1 hard-mining heuristic (ablation) | Fast R-CNN, VGG16 · VOC07 test | mAP: 67.2 -> 67.5 | No clear effect |
| Use one image per batch instead of two (ablation) | Fast R-CNN, VGG16 · VOC07 test | mAP: 67.2 -> 66.3 | Hurt |
| Use one image per batch with OHEM instead of two (ablation) | Fast R-CNN, VGG16 · VOC07 test | mAP: 69.9 -> 69.7 | No clear effect |
Closest papers by meaning
- Towards Human-Machine Cooperation: Self-Supervised Sample Mining for Object Detection (CVPR 2018 · similarity 0.88)
- Object Instance Mining for Weakly Supervised Object Detection (AAAI 2020 · similarity 0.87)
- R-FCN: Object Detection via Region-based Fully Convolutional Networks (NeurIPS 2016 · similarity 0.86)
- OW-OVD: Unified Open World and Open Vocabulary Object Detection (CVPR 2025 · similarity 0.85)
- Plug and Play Active Learning for Object Detection (CVPR 2024 · similarity 0.85)
- Search and Detect: Training-Free Long Tail Object Detection via Web-Image Retrieval (CVPR 2025 · similarity 0.85)
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.