Research map / Detection and segmentation / CNN features
PIoU Loss: Towards Accurate Oriented Object Detection in Complex Environments
Zhiming Chen, Kean Chen, Weiyao Lin, John See, Hui Yu, Yan Ke, Cong Yang
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 |
|---|---|---|---|
| replace SmoothL1 with PIoU loss (ablation) | RefineDet (VGG, FPN) · DOTA | AP: 51.46 -> 54.24 | Improved |
| replace L1 with PIoU loss (ablation) | RefineDet (VGG, FPN) · DOTA | AP: 50.66 -> 54.24 | Improved |
| replace L2 with PIoU loss (ablation) | RefineDet (VGG, FPN) · DOTA | AP: 49.70 -> 54.24 | Improved |
| use the Hard PIoU approximation instead of PIoU (ablation) | RefineDet · DOTA | AP: 54.24 -> 53.37 | Mixed |
| sensitivity factor k=10 instead of 5 (ablation) | PIoU loss · DOTA | AP: 46.88 -> 54.24 | Improved |
| sensitivity factor k=10 instead of 15 (ablation) | PIoU loss · DOTA | AP: 53.41 -> 54.24 | Improved |
| add PIoU loss to RefineDet-OBB | RefineDet-OBB ResNet-50 · Retail50K | AP: 53.96 -> 61.78 | Improved |
| add PIoU loss to CenterNet-OBB | CenterNet-OBB ResNet18 · Retail50K | AP: 54.44 -> 61.02 | Improved |
| add PIoU loss to CenterNet-OBB | CenterNet-OBB ResNet18 · HRSC2016 | mAP: 67.73 -> 78.54 | Improved |
| add PIoU loss to CenterNet-OBB | CenterNet-OBB DLA-34 · HRSC2016 | mAP: 87.98 -> 89.20 | Improved |
| add PIoU loss to RefineDet | RefineDet VGG16 · DOTA | mAP: 50.9 -> 52.5 | Improved |
| PIoU versus GIoU loss for horizontal boxes | SSD · PASCAL VOC2007 | AP: 49.9 -> 50.3 | Improved |
Closest papers by meaning
- Distance-IoU Loss: Faster and Better Learning for Bounding Box Regression (AAAI 2020 · similarity 0.90)
- Rethinking Boundary Discontinuity Problem for Oriented Object Detection (CVPR 2024 · similarity 0.87)
- Theoretically Achieving Continuous Representation of Oriented Bounding Boxes (CVPR 2024 · similarity 0.86)
- PointOBB: Learning Oriented Object Detection via Single Point Supervision (CVPR 2024 · similarity 0.86)
- ABBSPO: Adaptive Bounding Box Scaling and Symmetric Prior based Orientation Prediction for Detecting Aerial Image Objects (CVPR 2025 · similarity 0.85)
- GauCho: Gaussian Distributions with Cholesky Decomposition for Oriented Object Detection (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.
