Generalized Compositional Zero-Shot Learning on MIT-States (test)
6.5AUC (1)CGE
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| CGEFeature Extraction Strategy=end-to-end2021.02 | 6.5 | — | — | 32.8 | 28 | 21.4 | 30.1 | 34.7 | |
| CGEffFeature Extraction Strategy=fixed Resnet182021.02 | 5.1 | — | — | 28.7 | 25.3 | 17.2 | 27.9 | 32 | |
| SymNetBackbone=ResNet-182020.04 | 3 | 7.6 | 12.3 | 24.4 | 25.2 | 16.1 | — | — | |
| SymNetFeature Extraction Strategy=fixed2021.02 | 3 | — | — | 24.4 | 25.2 | 16.1 | 26.3 | 28.3 | |
| TMNBackbone=ResNet-182020.04 | 2.9 | 7.1 | 11.5 | 20.2 | 20.1 | 13 | — | — | |
| TMNFeature Extraction Strategy=fixed2021.02 | 2.9 | — | — | 20.2 | 20.1 | 13 | 23.3 | 26.5 | |
| Red WineBackbone=ResNet-182020.04 | 2.4 | 5.7 | 9.3 | 20.7 | 17.9 | 11.6 | — | — | |
| GenModelBackbone=ResNet-182020.04 | 2.3 | 5.7 | 8.8 | 24.8 | 13.4 | 11.2 | — | — | |
| LabelEmbed+Backbone=ResNet-182020.04 | 2 | 5.6 | 9.4 | 15 | 20.1 | 10.7 | — | — | |
| LE+Feature Extraction Strategy=fixed2021.02 | 2 | — | — | 15 | 20.1 | 10.7 | 23.5 | 26.3 | |
| AttrOperatorBackbone=ResNet-182020.04 | 1.6 | 4.7 | 7.6 | 14.3 | 17.4 | 9.9 | — | — | |
| AttOpFeature Extraction Strategy=fixed2021.02 | 1.6 | — | — | 14.3 | 17.4 | 9.9 | 21.1 | 23.6 |