Research map / Detection and segmentation / CNN features
R3Det: Refined Single-Stage Detector with Feature Refinement for Rotating Object
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 |
|---|---|---|---|
| Add feature refinement with box filtering and feature reconstruction (ablation) | R³Det* · DOTA | mAP: 63.52 -> 66.31 | Helped |
| Add feature refinement (ablation) | R³Det* · ICDAR2015 | Hmean: 83.27 -> 84.96 (+1.69) | Helped |
| Add feature refinement (ablation) | R³Det* · HRSC2016 (2012 metric) | mAP: 94.98 -> 96.01 (+1.03) | Helped |
| Add feature refinement (ablation) | R³Det* · UCAS-AOD | mAP: 95.03 -> 96.17 (+1.14) | Helped |
| Use three refinement stages rather than two (ablation) | R³Det · DOTA | mAP: 66.31 -> 67.29 | Helped |
| Use four refinement stages rather than three (ablation) | R³Det · DOTA | mAP: 67.29 -> 67.02 | Hurt |
| Ensemble outputs from refinement stages 2–3 (ablation) | R³Det · DOTA | mAP: 67.66 vs 67.29 | Helped |
| Use the proposed location-sensitive interpolation instead of randomized or quantized feature sampling (ablation) | FRM · DOTA | mAP: 66.31 vs 64.28; 64.37; 64.02; 64.19 | Helped |
| Use FRM instead of deformable alignment | Feature alignment · DOTA | mAP: 63.56 -> 66.31 | Helped |
| Use 1-SkewIoU loss instead of baseline loss (ablation) | RetinaNet-H · DOTA | mAP: 62.79 -> 65.06 (+2.27) | Helped |
| Use exp(1-SkewIoU)-1 loss instead of baseline loss (ablation) | RetinaNet-H · DOTA | mAP: 62.79 -> 65.34 (+2.55) | Helped |
| Use exp(1-SkewIoU)-1 rather than 1-SkewIoU (ablation) | R³Det† · DOTA | mAP: 68.97 -> 69.50 | Helped |
Closest papers by meaning
- ReDet: A Rotation-Equivariant Detector for Aerial Object Detection (CVPR 2021 · similarity 0.89)
- FRED: Towards a Full Rotation-Equivariance in Aerial Image Object Detection (AAAI 2024 · similarity 0.86)
- R2Det: Exploring Relaxed Rotation Equivariance in 2D Object Detection (ICLR 2025 · similarity 0.85)
- Rethinking Boundary Discontinuity Problem for Oriented Object Detection (CVPR 2024 · similarity 0.84)
- ReDiffDet: Rotation-equivariant Diffusion Model for Oriented Object Detection (CVPR 2025 · similarity 0.84)
- Leveraging 3D Geometric Priors in 2D Rotation Symmetry Detection (CVPR 2025 · similarity 0.83)
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.