Object Detection on Cross-Domain 10-Shot (ArTaxOr, Clipart1k, DIOR, DeepFish, NEU-DET, UODD)
54.98mAP (ArTaxOr)Ours
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| OursBackbone=Swin-T, Shot count=52026.03 | 54.98 | 44.83 | 29.41 | 33.87 | 23.64 | 15.66 | 33.73 | |
| FRONTBackbone=ViT-B, Evaluation Protocol=10-shot2026.03 | 53.48 | 39.49 | 25.48 | 19.71 | 13.21 | 5.78 | 26.19 | |
| ScratchBackbone=ViT-B, Evaluation Protocol=10-shot2026.03 | 51.42 | 37.59 | 24.5 | 19.22 | 12.99 | 5.81 | 25.26 | |
| GLIPBackbone=Swin-T, Shot count=52026.03 | 49.11 | 39.28 | 28.39 | 28.4 | 19.55 | 11.4 | 29.36 | |
| VFM-MoEBackbone=Swin-T, Shot count=5, Implementation=Reimplemented2026.03 | 48.23 | 39.13 | 26.34 | 27.58 | 18.67 | 10.73 | 28.45 | |
| CD-ViTOBackbone=Swin-T, Shot count=5, Implementation=Reimplemented2026.03 | 48.12 | 40.28 | 27.05 | 28.26 | 16.84 | 9.24 | 28.3 | |
| CD-ViTOBackbone=ViT-L/14, Shot count=5, Source=Reported in CD-ViTO [5]2026.03 | 47.9 | 41.1 | 26.9 | 22.3 | 11.4 | 6.8 | 26.07 | |
| Domain-RAGBackbone=Swin-T, Shot count=5, Implementation=Reimplemented2026.03 | 46.13 | 41.47 | 25.23 | 27.67 | 17.53 | 9.87 | 27.98 | |
| FRONTBackbone=ViT-S, Evaluation Protocol=10-shot2026.03 | 42.36 | 29.74 | 23.33 | 20.73 | 12.71 | 5.09 | 22.33 | |
| ScratchBackbone=ViT-S, Evaluation Protocol=10-shot2026.03 | 41.97 | 27.7 | 21.4 | 20 | 12.57 | 5.15 | 21.47 | |
| DE-ViTBackbone=Swin-T, Shot count=5, Implementation=Reimplemented2026.03 | 38.15 | 38.27 | 23.52 | 25.89 | 11.5 | 6.79 | 24.02 | |
| DE-ViTBackbone=ViT-L/14, Shot count=5, Source=Reported in CD-ViTO [5]2026.03 | 38 | 38.1 | 23.4 | 21.2 | 7.8 | 5 | 22.25 | |
| ViTDetBackbone=Swin-T, Shot count=5, Implementation=Reimplemented2026.03 | 21.05 | 23.45 | 23.45 | 11.03 | 19.96 | 15.07 | 19 | |
| ViTDetBackbone=ViT-B/14, Shot count=5, Source=Reported in CD-ViTO [5]2026.03 | 20.9 | 23.3 | 23.3 | 9 | 13.5 | 11.1 | 16.85 | |
| DistillBackbone=ResNet50, Shot count=5, Source=Reported in CD-ViTO [5]2026.03 | 12.5 | 23.3 | 19.1 | 15.5 | 16 | 12.2 | 16.43 | |
| DeticBackbone=Swin-T, Shot count=5, Implementation=Reimplemented2026.03 | 10.63 | 20.31 | 14.78 | 17.46 | 20.8 | 14.07 | 16.34 | |
| DeticBackbone=ViT-L/14, Shot count=5, Source=Reported in CD-ViTO [5]2026.03 | 8.7 | 20.2 | 12.1 | 14.3 | 14.1 | 10.4 | 13.3 |