Research map / Detection and segmentation / Weakly supervised detection and segmentation
Point-to-Box Network for Accurate Object Detection via Single Point Supervision
Pengfei Chen, Xuehui Yu, Xumeng Han, Najmul Hassan, Kai Wang, Jiachen Li, Jian Zhao, Humphrey Shi, Zhenjun Han, Qixiang Ye
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 negative sampling and negative loss L_neg in PBR (ablation) | P2BNet · COCO-17 val | AP: 12.7 -> 21.7 | Improved |
| add one PBR refinement stage after CBP (ablation) | P2BNet · COCO-17 val | AP: 13.7 -> 21.7 | Improved |
| use two PBR iterations instead of one (ablation) | P2BNet · COCO-17 val | AP: 21.7 -> 21.9 | No clear effect |
| use three PBR iterations instead of one (ablation) | P2BNet · COCO-17 val | AP: 21.7 -> 21.3 | Hurt |
| add center jitter to PBR proposals (ablation) | P2BNet · COCO-17 val | AP: 14.2 -> 21.7 | Improved |
| use balanced proposal sampling per object (ablation) | P2BNet · COCO-17 val | AP: 12.9 -> 21.7 | Improved |
| use L_mil1 instead of treating all bag proposals as positive (L_pos) (ablation) | P2BNet CBP · COCO-17 val | AP: 2.9 -> 13.7 | Improved |
| add L_pseudo using CBP pseudo boxes as positives (ablation) | P2BNet · COCO-17 val | AP: 21.7 -> 18.5 | Hurt |
| merge boxes with bag score instead of classification score (ablation) | P2BNet · COCO-17 val | AP: 17.4 -> 21.7 | Improved |
| increase top-k from 1 to 4 in box merging (ablation) | P2BNet · COCO-17 val | AP: 12.2 -> 22.1 | Improved |
| use ResNet-101 instead of ResNet-50 backbone in P2BNet (ablation) | P2BNet · COCO-17 val | AP: 21.7 -> 20.8 | Hurt |
| replace OTSP-based UFO2 with P2BNet-FR (QC points) | P2BNet-FR · COCO-17 val | AP50: 28.9 -> 47.3 | Improved |
Closest papers by meaning
- P2Object: Single Point Supervised Object Detection and Instance Segmentation (IJCV 2025 · similarity 0.94)
- Point2RBox: Combine Knowledge from Synthetic Visual Patterns for End-to-end Oriented Object Detection with Single Point Supervision (CVPR 2024 · similarity 0.88)
- PointOBB-v2: Towards Simpler, Faster, and Stronger Single Point Supervised Oriented Object Detection (ICLR 2025 · similarity 0.88)
- PointOBB: Learning Oriented Object Detection via Single Point Supervision (CVPR 2024 · similarity 0.87)
- Point2RBox-v2: Rethinking Point-supervised Oriented Object Detection with Spatial Layout Among Instances (CVPR 2025 · similarity 0.86)
- PointSR: Self-Regularized Point Supervision for Drone-View Object Detection (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.
