Expression Classification on AffectNet (test)
63.11Total AccuracyDMUE
Evaluation results
| Method | Links | ||
|---|---|---|---|
| DMUEBackbone=ResNet-50IBN2021.04 | 63.11 | — | |
| DMUEBackbone=ResNet-182021.04 | 62.84 | — | |
| CAKETraining Set=AffectNet and RAF-DB2021.04 | 61.7 | — | |
| FMPNSetting=image-based2019.02 | 61.52 | — | |
| CNN (baseline)Setting=image-based2019.02 | 60.86 | — | |
| SCNNumber of Classes=7 classes2021.04 | 60.23 | — | |
| CNN (no lg)Setting=image-based2019.02 | 60.01 | — | |
| VGG-FACE trained using the proposed approachBackbone=VGG-FACE, Training Method=Facial affect synthesis framework2018.11 | 60 | 59 | |
| RAN2021.04 | 59.5 | — | |
| AlexNetBackbone=AlexNet2018.11 | 58 | 58 | |
| IPA2LTTraining Set=AffectNet and RAF-DB2021.04 | 55.71 | — | |
| VGG-FACE baselineBackbone=VGG-FACE, Training Data=AffectNet training set2018.11 | 52 | 51 | |
| Upsample2021.04 | 47.01 | — |