Research map / Detection and segmentation / Open-set and few-shot detection
Free-Lunch Augmentation by Revisiting Diffusion-Based Data Generation for Cross-Domain Few-Shot Object Detection
Zijian Zhuang, Yixiong Zou, Yuhua Li, Ruixuan Li
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 SITN synthesis to CD-ViTO | CD-ViTO (ViT-L) · CDFSOD 1-shot average | mAP: 13.9 -> 17.0 | Improved |
| add SITN to GroundDINO | GroundDINO (Swin-B) · CDFSOD 1-shot average | mAP: 24.0 -> 34.3 | Improved |
| add SITN to ViTDeT-FT on ArTaxOr and DIOR | ViTDeT-FT · CDFSOD 1-shot ArTaxOr | mAP: 5.9 -> 5.4 (ArTaxOr); average 5.4 -> 6.5 | Mixed |
| add SITN to ETS | ETS (Swin-B) · CDFSOD 1-shot average | mAP: 21.9 -> 32.1 | Improved |
| add background inpainting only (ablation) | CD-ViTO · ArTaxOr 1-shot | mAP: 21.0 -> 24.9 | Improved |
| combine foreground and background without selection (ablation) | CD-ViTO · Deepfish 1-shot | mAP: 20.3 -> 13.5 | Hurt |
| add selection module (ablation) | SITN · UODD 1-shot | mAP: 2.1 -> 5.6 | Improved |
| use foreground-only inpainting (ablation) | CD-ViTO · Deepfish 1-shot | mAP: 20.3 -> 18.3 | Hurt |
| use standard diffusion generators (Diffmix, IP-Adapter, etc.) instead of SITN (ablation) | CD-ViTO · CDFSOD 1-shot average | mAP: 13.9 (baseline) -> 4.8 (Diffmix) | Hurt |
| apply Mixup augmentation (ablation) | CD-ViTO · CDFSOD 1-shot average | mAP: 13.9 -> 8.7 | Hurt |
| use Nucleus-Image generator inside SITN (ablation) | SITN · CDFSOD average | mAP: 34.3 -> 32.1 | No clear effect |
| apply SITN to FPTrans for segmentation | FPTrans · CDFSS average | mIoU: 62.2 -> 66.7 | Improved |
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.
- DomainRAG — Domain-RAG: Retrieval-Guided Compositional Image Generation for Cross-Domain Few-Shot Object Detection (NeurIPS 2025) · 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
- Domain-RAG: Retrieval-Guided Compositional Image Generation for Cross-Domain Few-Shot Object Detection (NeurIPS 2025 · similarity 0.89)
- Multi-Perspective Data Augmentation for Few-shot Object Detection (ICLR 2025 · similarity 0.88)
- The Devil is in Low-Level Features for Cross-Domain Few-Shot Segmentation (CVPR 2025 · similarity 0.88)
- Generalized Diffusion Detector: Mining Robust Features from Diffusion Models for Domain-Generalized Detection (CVPR 2025 · similarity 0.88)
- Adapter Naturally Serves as Decoupler for Cross-Domain Few-Shot Semantic Segmentation (ICML 2025 · similarity 0.87)
- Adapt Before Comparison: A New Perspective on Cross-Domain Few-Shot Segmentation (CVPR 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.
