Table 5 Classification accuracy (%) on the VisDA-2017 dataset for various domain adaptation methods. The best results are highlighted in bold.
Method | Backbone | airplane | bicycle | bus | car | horse | knife | motorcycle | person | plant | skateboard | train | truck | AVG |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Baseline | ResNet-101 | 55.1 | 53.3 | 61.9 | 59.1 | 80.6 | 17.9 | 79.7 | 31.2 | 81.0 | 26.5 | 73.5 | 8.5 | 52.36 |
CGDMÂ 21 | 93.4 | 82.7 | 73.2 | 68.4 | 92.9 | 94.5 | 88.7 | 82.1 | 93.4 | 82.5 | 86.8 | 49.2 | 82.32 | |
SDATÂ 39 | 95.8 | 85.5 | 76.9 | 69.0 | 93.5 | 97.4 | 88.5 | 78.2 | 93.1 | 91.6 | 86.3 | 55.3 | 84.26 | |
CLIPÂ 14 | 98.2 | 83.9 | 90.5 | 73.5 | 97.2 | 84.0 | 95.3 | 65.7 | 79.4 | 89.9 | 91.8 | 63.3 | 84.39 | |
Ours | 96.7 | 83.9 | 80.9 | 76.2 | 97.0 | 98.1 | 90.0 | 84.0 | 96.5 | 94.7 | 88.7 | 57.5 | 87.01 | |
Baseline | ViT | 98.2 | 73.0 | 82.5 | 62.0 | 97.3 | 63.5 | 96.5 | 29.8 | 68.7 | 86.7 | 96.7 | 23.7 | 73.22 |
TVTÂ 24 | 92.9 | 85.6 | 77.5 | 60.5 | 93.6 | 98.2 | 89.4 | 76.4 | 93.6 | 92.0 | 91.7 | 55.7 | 83.92 | |
PMTrans 23 | 98.9 | 93.7 | 84.5 | 73.3 | 99.0 | 98.0 | 96.2 | 67.8 | 94.2 | 98.4 | 96.6 | 49.0 | 87.47 | |
CLIPÂ 14 | 99.3 | 91.7 | 93.9 | 74.3 | 98.4 | 94.3 | 90.3 | 78.2 | 78.3 | 97.3 | 95.2 | 64.8 | 88.00 | |
SDATÂ 39 | 98.4 | 90.9 | 85.4 | 82.1 | 98.5 | 97.6 | 96.3 | 86.1 | 96.2 | 96.7 | 92.9 | 56.8 | 89.83 | |
CMKDÂ 13 | 99.4 | 94.6 | 91.5 | 78.9 | 98.7 | 97.3 | 93.3 | 81.3 | 91.8 | 97.9 | 96.9 | 61.7 | 90.28 | |
Ours | 98.4 | 94.8 | 87.2 | 88.7 | 100.0 | 100.0 | 97.6 | 89.0 | 97.8 | 98.3 | 92.0 | 68.5 | 92.68 |
