Research map / Detection and segmentation / CNN features
Learning High-Precision Bounding Box for Rotated Object Detection via Kullback-Leibler Divergence
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 |
|---|---|---|---|
| replace Smooth L1 with KLD regression loss | RetinaNet · HRSC2016 | AP75: 48.42 -> 72.39 | Helped |
| replace Smooth L1 with KLD regression loss | R3Det · HRSC2016 | AP75: +33.96 | Helped |
| replace GWD with KLD loss | RetinaNet · DOTA-v1.0 | AP50: 68.93 -> 71.28 | Helped |
| replace Smooth L1 with KLD | RetinaNet · ICDAR2015 (R+F) | Hmean85: 11.59 -> 11.09 | Hurt |
| replace Smooth L1 with KLD | RetinaNet · MLT | Hmean50: +9.17 | Helped |
| replace Smooth L1/GIoU with KLD for horizontal detection | RetinaNet · MS COCO | AP: 37.2 -> 38.0 | Helped |
| replace GIoU with KLD | Faster RCNN · MS COCO | AP: 38.3 -> 38.2 | No clear effect |
| add nonlinear transform log(D+1) to raw KLD (ablation) | KLD loss · HRSC2016 | AP50: 0.20 -> 83.23 | Helped |
| add normalization 1-1/(tau+f(D)) with tau=1 (ablation) | KLD loss · HRSC2016 | AP50: 83.23 -> 85.25 | Helped |
| increase tau from 1 to 5 (ablation) | KLD loss · HRSC2016 | AP50: 85.25 -> 73.44 | Hurt |
| use symmetric/other KLD variants instead of forward KL (ablation) | KLD loss · DOTA-v1.0 | AP50: 70.17 -> 70.71 | No clear effect |
| apply the same normalization to Smooth L1 (ablation) | Smooth L1 · HRSC2016 | Hmean50: 84.80 -> 78.99 | Hurt |
Closest papers by meaning
- Rethinking Boundary Discontinuity Problem for Oriented Object Detection (CVPR 2024 · similarity 0.89)
- ReDiffDet: Rotation-equivariant Diffusion Model for Oriented Object Detection (CVPR 2025 · similarity 0.88)
- FRED: Towards a Full Rotation-Equivariance in Aerial Image Object Detection (AAAI 2024 · similarity 0.88)
- ABBSPO: Adaptive Bounding Box Scaling and Symmetric Prior based Orientation Prediction for Detecting Aerial Image Objects (CVPR 2025 · similarity 0.87)
- R2Det: Exploring Relaxed Rotation Equivariance in 2D Object Detection (ICLR 2025 · similarity 0.86)
- Theoretically Achieving Continuous Representation of Oriented Bounding Boxes (CVPR 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.