Research map / Detection and segmentation / CNN features
On the Arbitrary-Oriented Object Detection: Classification Based Approaches Revisited
Xue Yang, Junchi Yan
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 angle regression with CSL classification | RetinaNet · DOTA-v1.0 val | AP50:95: 31.49 -> 35.04 | Improved |
| replace angle regression with BCL classification | RetinaNet · DOTA-v1.0 val | AP50:95: 31.49 -> 36.71 | Improved |
| use sin/cos indirect regression instead of direct regression | RetinaNet · DOTA-v1.0 val | AP50:95: 31.49 -> 33.19 | Improved |
| replace regression with CSL on RetinaNet-H angle [-90,90) | RetinaNet-H · DOTA-v1.0 test | mAP: 64.17 -> 67.38 | Improved |
| replace regression with CSL on FPN-H angle [-90,90) | FPN-H · DOTA-v1.0 test | mAP: 69.45 -> 70.92 | Improved |
| replace regression with BCL on R3Det | R3Det · DOTA-v1.0 test | mAP: 70.66 -> 71.21 | Improved |
| use Gaussian window instead of pulse window (ablation) | RetinaNet-CSL · DOTA-v1.0 test | mAP: 42.06 -> 65.69 | Improved |
| use Gaussian window instead of triangle window (ablation) | RetinaNet-CSL · DOTA-v1.0 test | mAP: 57.94 -> 65.69 | Improved |
| increase window radius from 0 to 6 (ablation) | RetinaNet-CSL · DOTA-v1.0 test | mAP: 40.78 -> 65.69 | Improved |
| increase window radius from 6 to 8 (ablation) | RetinaNet-CSL · DOTA-v1.0 test | mAP: 65.69 -> 63.99 | Hurt |
| coarsen angle granularity from omega=10 to omega=30 (ablation) | RetinaNet-CSL · DOTA-v1.0 test | mAP: 67.38 -> 40.81 | Hurt |
| add angle fine-tuning to CSL (omega=10) (ablation) | CSL · HRSC2016 | AP75: 16.89 -> 32.57 | Improved |
Closest papers by meaning
- Arbitrary-Oriented Object Detection with Circular Smooth Label (ECCV 2020 · similarity 0.95)
- Dense Label Encoding for Boundary Discontinuity Free Rotation Detection (CVPR 2021 · similarity 0.89)
- On Boundary Discontinuity in Angle Regression Based Arbitrary Oriented Object Detection (TPAMI 2024 · similarity 0.88)
- Rethinking Boundary Discontinuity Problem for Oriented Object Detection (CVPR 2024 · similarity 0.88)
- FRED: Towards a Full Rotation-Equivariance in Aerial Image Object Detection (AAAI 2024 · similarity 0.86)
- PointOBB: Learning Oriented Object Detection via Single Point Supervision (CVPR 2024 · 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.
