Research map / Detection and segmentation / CNN features
Oriented Tiny Object Detection: A Dataset, Benchmark, and Dynamic Unbiased Learning
Chang Xu, Ruixiang Zhang, Wen Tao Yang, Haoran Zhu, Fang Xu, Jian Ding, Gui-Song Xia
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 |
|---|---|---|---|
| add MPS (posterior re-ranked candidates) to CPS (ablation) | DCFL on RetinaNet-O · DOTA-v2 val | mAP50: 51.70 -> 53.41 | Improved |
| add DGMM on top of CPS+MPS (ablation) | DCFL on RetinaNet-O · DOTA-v2 val | mAP50: 53.41 -> 59.15 | Improved |
| add MPS on top of CPS+DGMM (ablation) | DCFL on RetinaNet-O · DOTA-v2 val | mAP50: 57.20 -> 59.15 | Improved |
| use cross-FPN-layer CPS instead of all-FPN-layer candidates (ablation) | DCFL CPS · DOTA-v2 val | mAP50: 50.12 -> 59.15 | Improved |
| use cross-FPN-layer CPS instead of single-FPN-layer candidates (ablation) | DCFL CPS · DOTA-v2 val | mAP50: 56.72 -> 59.15 | Improved |
| use GJSD instead of KLD as CPS measurement (ablation) | DCFL CPS · DOTA-v2 val | mAP50: 57.82 -> 59.15 | Improved |
| add regression-guided classification (Guiding) in PCB (ablation) | DCFL PCB · DOTA-v2 val | mAP50: 58.71 -> 59.15 | Improved |
| vary DGMM threshold g from 1.2 to 0.8 (ablation) | DCFL · DOTA-v2 val | mAP50: 57.91 -> 59.15 | Improved |
| use Swin-T backbone instead of ResNet-50 | Oriented R-CNN · AI-TOD-R | AP: 11.2 -> 12.0 | Improved |
| use deeper ResNet-101 backbone instead of ResNet-50 | Oriented R-CNN · AI-TOD-R | AP: 11.2 -> 11.2 | No clear effect |
| add DCFL to RetinaNet-O (1x) | RetinaNet-O · AI-TOD-R | AP: 7.3 -> 12.3 | Improved |
| add DCFL to Oriented R-CNN (1x) | Oriented R-CNN · AI-TOD-R | AP: 11.2 -> 15.7 | Improved |
Closest papers by meaning
- Dynamic Coarse-To-Fine Learning for Oriented Tiny Object Detection (CVPR 2023 · similarity 0.92)
- Towards Large-Scale Small Object Detection: Survey and Benchmarks (TPAMI 2023 · similarity 0.91)
- Deep Learning for Generic Object Detection: A Survey (IJCV 2019 · similarity 0.89)
- Frustratingly Simple Few-Shot Object Detection (ICML 2020 · similarity 0.89)
- "A Simple Approach and Benchmark for 21,000-Category Object Detection" (ECCV 2022 · similarity 0.89)
- Roboflow100-VL: A Multi-Domain Object Detection Benchmark for Vision-Language Models (NeurIPS 2025 · similarity 0.88)
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.
