Research map / Detection and segmentation / CNN features
GRA: Detecting Oriented Objects through Group-wise Rotating and Attention
Jiangshan Wang, Yifan Pu, Yizeng Han, Jiayi Guo, Yiru Wang, Xiu Li, Gao Huang
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 ARC with GRA (group-wise rotating plus attention) in ResNet-50 | Oriented R-CNN on DOTA-v1.0 · DOTA-v1.0 | mAP: 77.35 -> 77.63 with backbone params 57.20M -> 23.79M | Improved |
| replace ARC with GRA across five detectors | Rotated RetinaNet, R3Det, S2ANet, Rotated Faster RCNN, CFA · DOTA-v1.0 | mAP: e.g. Rotated RetinaNet 71.45 -> 72.19 | Improved |
| replace ARC with GRA | Oriented R-CNN on DOTA-v2.0 · DOTA-v2.0 | mAP: 55.91 -> 56.63 | Improved |
| train GRA for 40 epochs | Oriented R-CNN R50_GRA · DOTA-v2.0 | mAP: 56.63 -> 57.95 | Improved |
| add GRA in place of ResNet 3x3 convs | HRSC2016 detectors · HRSC2016 | mAP: Oriented R-CNN 70.55 -> 72.59 | Improved |
| add Group-wise Rotating (ablation) | Oriented R-CNN R50 · DOTA-v1.0 | mAP: 75.81 -> 76.73 | Improved |
| add learnable group scale factor lambda (ablation) | Group-wise Rotating · DOTA-v1.0 | mAP: 76.73 -> 77.25 | Improved |
| add Group-wise Attention (ablation) | GRA · DOTA-v1.0 | mAP: 77.25 -> 77.63 | Improved |
| increase number of groups n from 2 to 32 (ablation) | GRA · DOTA-v1.0 | mAP: 76.82 -> 77.63 | Improved |
| train only GRA on frozen public pretrained ResNet instead of no pretraining (ablation) | Oriented R-CNN R50_GRA · DOTA-v1.0 | mAP: 35.51 -> 77.39 | Improved |
Closest papers by meaning
- Adaptive Rotated Convolution for Rotated Object Detection (ICCV 2023 · similarity 0.91)
- SOOD++: Leveraging Unlabeled Data to Boost Oriented Object Detection (TPAMI 2025 · similarity 0.89)
- FRED: Towards a Full Rotation-Equivariance in Aerial Image Object Detection (AAAI 2024 · similarity 0.87)
- PointOBB-v2: Towards Simpler, Faster, and Stronger Single Point Supervised Oriented Object Detection (ICLR 2025 · similarity 0.87)
- R2Det: Exploring Relaxed Rotation Equivariance in 2D Object Detection (ICLR 2025 · similarity 0.87)
- ReDiffDet: Rotation-equivariant Diffusion Model for Oriented 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.
