Image Classification on DomainNet 1.0 (test)
0.8292Accuracy (Real)LP-FT
Evaluation results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| LP-FTBackbone=ResNet50, Pre-training=MOCO-V3, Data Percentage=100%2023.10 | 0.8292 | 0.345 | 0.4542 | 0.2012 | 0.4711 | 0.3679 | 1.13 | 6.72 | |
| TPGMBackbone=ResNet50, Pre-training=MOCO-V3, Data Percentage=100%, Time (s/it)=0.82023.10 | 0.8266 | 0.3535 | 0.462 | 0.2013 | 0.4575 | 0.3686 | 0.82 | 6.91 | |
| FTPBackbone=ResNet50, Pre-training=MOCO-V3, Data Percentage=100%, Time (s/it)=0.512023.10 | 0.8217 | 0.3626 | 0.4658 | 0.2067 | 0.4697 | 0.3762 | 0.22 | 9.13 | |
| Vanilla FTBackbone=ResNet50, Pre-training=MOCO-V3, Data Percentage=100%, Time (s/it)=0.352023.10 | 0.8199 | 0.3152 | 0.4289 | 0.1851 | 0.4498 | 0.3447 | 0 | 0 | |
| MARS-SPBackbone=ResNet50, Pre-training=MOCO-V3, Data Percentage=100%, Time (s/it)=0.432023.10 | 0.8189 | 0.3444 | 0.4505 | 0.1997 | 0.4636 | 0.3645 | -0.13 | 5.74 | |
| L2-SPBackbone=ResNet50, Pre-training=MOCO-V3, Data Percentage=100%, Time (s/it)=0.462023.10 | 0.8151 | 0.3491 | 0.4576 | 0.1897 | 0.4529 | 0.3623 | -0.59 | 5.09 | |
| Partial FusionBackbone=ResNet50, Pre-training=MOCO-V3, Data Percentage=100%, Time (s/it)=0.212023.10 | 0.7827 | 0.2772 | 0.3974 | 0.1556 | 0.3818 | 0.303 | -4.55 | -12.11 | |
| Linear Prob.Backbone=ResNet50, Pre-training=MOCO-V3, Data Percentage=100%, Time (s/it)=0.12023.10 | 0.7301 | 0.241 | 0.3956 | 0.1227 | 0.3038 | 0.2658 | -10.96 | -22.9 |