Research map / Detection and segmentation / Open-set and few-shot detection
AsyFOD: An Asymmetric Adaptation Paradigm for Few-Shot Domain Adaptive Object Detection
Yipeng Gao, Kun-Yu Lin, Junkai Yan, Yaowei Wang, Wei-Shi Zheng
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 |
|---|---|---|---|
| Use AsyFOD instead of AcroFOD | YOLOv5 X detector · Cityscapes→Foggy Cityscapes | mAP50: 41.1±0.8 -> 44.3±1.0 | Improved |
| Use AsyFOD instead of AcroFOD | YOLOv5 X detector · Sim10K→Cityscapes | AP50: 62.5±1.6 -> 65.4±0.9 | Improved |
| Use AsyFOD instead of AcroFOD | YOLOv5 X detector · KITTI→Cityscapes | AP50: 62.6±2.1 -> 64.1±1.1 | Improved |
| Use AsyFOD instead of Pre+FT | YOLOv5 X detector · ViPeD→COCO | AP50: 43.2±0.8 -> 47.4±0.7 | Improved |
| Use AsyFOD instead of Pre+FT | YOLOv5 X detector · ViPeD→COCO | AP: 21.0±0.5 -> 23.1±0.5 | Improved |
| Add stop-gradient on target features during alignment (ablation) | AsyFOD alignment · Sim10K→Cityscapes | AP50: 62.2 -> 63.4 | Improved |
| Apply stop-gradient to source features rather than target features (ablation) | AsyFOD alignment · Sim10K→Cityscapes | AP50: 63.4 -> 61.7 | Hurt |
| Add target-similar source instances to augment target instances for alignment (ablation) | AsyFOD source-instance division · Sim10K→Cityscapes | AP50: 56.3±1.1 -> 57.4±0.5 | Improved |
| Estimate target-similar source instances using K-means rather than L2 distance (ablation) | AsyFOD · Sim10K→Cityscapes | AP50: 64.5±0.7 -> 65.4±0.9 | Improved |
| Estimate target-similar source instances using GMM rather than K-means (ablation) | AsyFOD · Cityscapes→Foggy Cityscapes | mAP50: 43.6±0.7 -> 44.3±1.0 | Improved |
| Use target-similar source and target instances for classification, while retaining all boxes for regression (ablation) | AsyFOD task-oriented supervised training · Sim10K→Cityscapes | AP50: 62.5±1.6 -> 64.2±1.1 | Improved |
| Restrict regression supervision to target-similar source and target instances (ablation) | AsyFOD task-oriented supervised training · Sim10K→Cityscapes | AP50: 64.2±1.1 -> 63.4±0.9 | Hurt |
Later papers that compared with this method
1 later accepted paper reports AsyFOD as a baseline in their results tables; 0 of them report a better number on most of the results they share with it.
- M3-UDA: A New Benchmark for Unsupervised Domain Adaptive Fetal Cardiac Structure Detection (CVPR 2024 · better on 0 of 1 shared result)
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- AcroFOD — AcroFOD: An Adaptive Method for Cross-Domain Few-Shot Object Detection (ECCV 2022) · better on 3 of 3 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- AcroFOD: An Adaptive Method for Cross-Domain Few-Shot Object Detection (ECCV 2022 · similarity 0.91)
- Domain Adaptive Object Detection via Asymmetric Tri-way Faster-RCNN (ECCV 2020 · similarity 0.91)
- DA-F2F: Domain-Adaptive Object Detection with Feature-to-Feature Modulation and Alignment (ECCV 2026 · similarity 0.90)
- Simplifying Source-Free Domain Adaptation for Object Detection: Effective Self-Training Strategies and Performance Insights (ECCV 2024 · similarity 0.90)
- Universal Domain Adaptive Object Detection via Dual Probabilistic Alignment (AAAI 2025 · similarity 0.89)
- CAT: Exploiting Inter-Class Dynamics for Domain Adaptive Object Detection (CVPR 2024 · 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.
