Research map / Detection and segmentation / CNN features
Dynamic Coarse-To-Fine Learning for Oriented Tiny 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 |
|---|---|---|---|
| Use DCFL instead of S²A-Net | One-stage oriented detector with ResNet-50 · DOTA-v2.0 OBB test | mAP: 49.86 -> 51.57 | Helped |
| Use DCFL instead of GGHL | One-stage oriented detector · DOTA-v1.0 OBB test | mAP: 73.98 -> 74.26 | Helped |
| Use DCFL instead of ReDet | Oriented detector · DOTA-v1.5 OBB test | mAP: 66.86 -> 67.37 | Helped |
| Use DCFL instead of Oriented Rep | Oriented detector · DIOR-R | mAP: 66.71 -> 66.80 | Helped |
| Add CPS and posterior-ranked MPS to the baseline (ablation) | RetinaNet-OBB · DOTA-v2.0 val | mAP: 51.70 -> 53.41 | Helped |
| Add DGMM with CPS (ablation) | RetinaNet-OBB · DOTA-v2.0 val | mAP: 51.70 -> 57.20 | Helped |
| Use CPS, MPS, and DGMM together (ablation) | RetinaNet-OBB · DOTA-v2.0 val | mAP: 51.70 -> 59.15 | Helped |
| Replace cross-layer GJSD CPS with all-layer Gaussian matching (ablation) | DCFL · DOTA-v2.0 val | mAP: 59.15 -> 50.12 | Hurt |
| Replace cross-layer GJSD CPS with single-layer Gaussian matching (ablation) | DCFL · DOTA-v2.0 val | mAP: 59.15 -> 56.72 | Hurt |
| Replace GJSD with KLD for cross-layer CPS (ablation) | DCFL · DOTA-v2.0 val | mAP: 59.15 -> 57.82 | Hurt |
| Replace GJSD with GWD for cross-layer CPS (ablation) | DCFL · DOTA-v2.0 val | mAP: 59.15 -> 58.55 | Hurt |
| Add DCN to the PCB (ablation) | DCFL · DOTA-v2.0 val | mAP: 58.07 -> 58.41 | Helped |
Closest papers by meaning
- One-to-Few Label Assignment for End-to-End Dense Detection (CVPR 2023 · similarity 0.86)
- Integrating Diverse Assignment Strategies into DETRs (AAAI 2026 · similarity 0.85)
- Multi-clue Consistency Learning to Bridge Gaps Between General and Oriented Object in Semi-supervised Detection (AAAI 2025 · similarity 0.84)
- Learning 1-Bit Tiny Object Detector with Discriminative Feature Refinement (ICML 2024 · similarity 0.84)
- Large Self-Supervised Models Bridge the Gap in Domain Adaptive Object Detection (CVPR 2025 · similarity 0.84)
- PointOBB-v2: Towards Simpler, Faster, and Stronger Single Point Supervised Oriented Object Detection (ICLR 2025 · similarity 0.84)
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.