Image Classification on Cross-Dataset Benchmark ResNet50 backbone (test)
19.89Accuracy (Aircraft)BCA
Evaluation Results
| Method | Links | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| BCABackbone=ResNet50, Training-free=true2025.07 | 19.89 | 89.7 | 42.12 | 66.3 | 85.58 | 63.38 | 63.51 | 58.13 | 48.58 | 77.19 | 61.44 | |
| DPEBackbone=ResNet50, Training-free=false2025.07 | 19.8 | 90.83 | 41.67 | 67.6 | 85.97 | 64.24 | 61.98 | 59.26 | 50.18 | 77.83 | 61.93 | |
| MS-TTABackbone=ResNet50, Training-free=true2025.07 | 19.23 | 88.52 | 47.61 | 68.94 | 86.02 | 63.05 | 64.68 | 59.61 | 43.97 | 78.85 | 62.05 | |
| BoostAdapterBackbone=ResNet50, Training-free=true2025.07 | 18.93 | 88.48 | 44.4 | 68.25 | 85.75 | 62.83 | 64.42 | 59.67 | 43.85 | 78.78 | 61.54 | |
| HisTPTBackbone=ResNet50, Training-free=false2025.07 | 18.1 | 87.2 | 42.5 | 67.6 | 84.9 | 63.5 | 64.1 | 61.3 | 41.3 | 81.3 | 61.18 | |
| TDABackbone=ResNet50, Training-free=true2025.07 | 17.61 | 89.7 | 42.11 | 68.74 | 86.18 | 62.53 | 64.18 | 57.78 | 43.74 | 77.75 | 61.03 | |
| DiffTPTBackbone=ResNet50, Training-free=false2025.07 | 17.6 | 86.89 | 41.04 | 63.53 | 83.4 | 62.72 | 62.97 | 60.71 | 40.72 | 79.21 | 59.88 | |
| TPTBackbone=ResNet50, Training-free=false2025.07 | 17.58 | 87.02 | 28.33 | 62.69 | 84.49 | 61.46 | 60.82 | 58.46 | 40.84 | 74.88 | 57.66 | |
| CLIPBackbone=ResNet50, Training-free=false2025.07 | 16.11 | 87.26 | 25.79 | 62.77 | 82.97 | 60.85 | 59.48 | 55.89 | 40.37 | 74.82 | 56.63 | |
| CoOpBackbone=ResNet50, Training-free=false2025.07 | 15.12 | 86.53 | 26.2 | 61.55 | 87 | 58.15 | 59.05 | 55.32 | 37.29 | 75.59 | 56.18 | |
| CoCoOpBackbone=ResNet50, Training-free=false2025.07 | 14.61 | 86.38 | 28.73 | 65.57 | 88.39 | 59.61 | 57.1 | 56.22 | 38.53 | 76.2 | 57.13 |