Research map / Detection and segmentation / Open-set and few-shot detection
AcroFOD: An Adaptive Method for Cross-Domain Few-Shot Object Detection
Yipeng Gao, Lingxiao Yang, Yunmu Huang, Song Xie, Shiyong Li, 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 |
|---|---|---|---|
| add adaptive optimization (filtering augmented samples) (ablation) | AcroFOD with augmentation · C->F | mAP50: 32.7 -> 41.1 | Improved |
| add multi-level domain-aware augmentation (ablation) | AcroFOD with optimization · C->F | mAP50: 26.4 -> 41.1 | Improved |
| add adaptive optimization (ablation) | AcroFOD with augmentation · S->C | AP50: 57.4 -> 62.5 | Improved |
| add augmentation (ablation) | AcroFOD with optimization · K->C | AP50: 51.9 -> 62.6 | Improved |
| add augmentation (ablation) | AcroFOD with optimization · V->O | AP50: 42.0 -> 45.8 | Improved |
| add adaptive optimization (ablation) | AcroFOD with augmentation · V->O | AP50: 41.4 -> 45.8 | Improved |
| use MMD instead of cosine distance for sample selection (ablation) | AcroFOD · S->C | AP50: 55.2 -> 55.6 | Improved |
| use MMD instead of cosine distance for sample selection (ablation) | AcroFOD · K->C | AP50: 51.4 -> 51.9 | Improved |
| add image-level domain-reallocation to domain-splice (ablation) | AcroFOD · C->F | mAP50: 27.5 -> 41.1 | Improved |
| add image-level domain-reallocation to domain-splice (ablation) | AcroFOD · S->C | AP50: 60.2 -> 62.3 | Improved |
| use proportional sampling of source and target data | YOLOv5 · S->C | AP50: 49.0 -> 50.2 | Improved |
| use AcroFOD instead of Pre+FT with 1% COCO person data | YOLOv5 · V->O 1% data | AP50: 50.9 -> 61.1 | Improved |
Later papers that compared with this method
2 later accepted papers report AcroFOD as a baseline in their results tables; 1 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)
- AsyFOD: An Asymmetric Adaptation Paradigm for Few-Shot Domain Adaptive Object Detection (CVPR 2023 · 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
- AsyFOD: An Asymmetric Adaptation Paradigm for Few-Shot Domain Adaptive Object Detection (CVPR 2023 · similarity 0.91)
- Cross-Domain Few-Shot Object Detection via Enhanced Open-Set Object Detector (ECCV 2024 · similarity 0.90)
- One-Shot Unsupervised Cross-Domain Detection (ECCV 2020 · similarity 0.90)
- Domain-RAG: Retrieval-Guided Compositional Image Generation for Cross-Domain Few-Shot Object Detection (NeurIPS 2025 · similarity 0.88)
- Multi-Perspective Data Augmentation for Few-shot Object Detection (ICLR 2025 · similarity 0.87)
- Fine-Grained Prototypes Distillation for Few-Shot Object Detection (AAAI 2024 · similarity 0.87)
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.
