Research map / Detection and segmentation / Weakly supervised detection and segmentation
SOOD: Towards Semi-Supervised Oriented Object Detection
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 RAW to the dense pseudo-label baseline (ablation) | rotated-FCOS · DOTA-v1.5-val, 10%, 20%, and 30% labeled | mAP: 47.24 -> 47.82; 54.07 -> 55.21; 57.74 -> 58.93 | Helped |
| Add GC to the dense pseudo-label baseline (ablation) | rotated-FCOS · DOTA-v1.5-val, 10%, 20%, and 30% labeled | mAP: 47.24 -> 47.71; 54.07 -> 54.72; 57.74 -> 58.70 | Helped |
| Add both RAW and GC (ablation) | rotated-FCOS dense pseudo-label baseline · DOTA-v1.5-val, 10%, 20%, and 30% labeled | mAP: 47.24 -> 48.63; 54.07 -> 55.58; 57.74 -> 59.23 | Helped |
| Set pseudo-label sampling ratio to 0.125, 0.5, or 1.0 instead of 0.25 (ablation) | SOOD · DOTA-v1.5-val, 10% labeled | mAP: 48.63 -> 48.27; 48.63 -> 47.91; 48.63 -> 47.69 | Hurt |
| Use both spatial distance and score difference in the GC cost map (ablation) | SOOD with RAW · DOTA-v1.5-val, 10% labeled | mAP: 47.82 -> 48.63 | Helped |
| Use only score difference in the GC cost map (ablation) | SOOD with RAW · DOTA-v1.5-val, 10% labeled | mAP: 47.82 -> 48.10 | Helped |
| Use only spatial distance in the GC cost map (ablation) | SOOD with RAW · DOTA-v1.5-val, 10% labeled | mAP: 47.82 -> 47.94 | Helped |
| Increase RAW alpha from 1 to 50 (ablation) | SOOD with GC · DOTA-v1.5-val, 10% labeled | mAP: 47.77 -> 48.63 | Helped |
| Increase RAW alpha from 50 to 100 (ablation) | SOOD with GC · DOTA-v1.5-val, 10% labeled | mAP: 48.63 -> 47.95 | Hurt |
| Add GC to Soft Teacher (ablation) | Soft Teacher · DOTA-v1.5-val, 10%, 20%, and 30% labeled | mAP: 48.46 -> 49.30; 54.89 -> 55.45; 57.83 -> 58.16 | Helped |
Closest papers by meaning
- Dense Learning Based Semi-Supervised Object Detection (CVPR 2022 · similarity 0.90)
- Multi-clue Consistency Learning to Bridge Gaps Between General and Oriented Object in Semi-supervised Detection (AAAI 2025 · similarity 0.90)
- Semi-Supervised Object Detection via Multi-Instance Alignment With Global Class Prototypes (CVPR 2022 · similarity 0.90)
- Sparse Semi-DETR: Sparse Learnable Queries for Semi-Supervised Object Detection (CVPR 2024 · similarity 0.87)
- Semi-supervised Open-World Object Detection (AAAI 2024 · similarity 0.87)
- Relational Matching for Weakly Semi-Supervised Oriented Object Detection (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.