Facial Expression Recognition on FERPlus (test)
0.9555AccuracyPAtt-Lite
Evaluation Results
| Method | Links | |
|---|---|---|
| PAtt-Lite# Params=1.10M2023.06 | 0.9555 | |
| CIAO# Params=17.9M2023.06 | 0.945 | |
| ARBEx# Params=-2023.06 | 0.9309 | |
| GReFELVenue=Current Paper2024.10 | 0.9309 | |
| POSTER++Venue='232024.10 | 0.9228 | |
| QCSPre-trained=AffectNet-82024.11 | 0.9185 | |
| LA-NetVenue=ICCV'232024.10 | 0.9178 | |
| POSTER# Params=71.8M2023.06 | 0.9162 | |
| POSTERVenue=IC-W'222024.10 | 0.9162 | |
| POSTERyear=20232024.11 | 0.9162 | |
| DCS2024.11 | 0.9141 | |
| QCS2024.11 | 0.9141 | |
| S2D w/o TMAPre-trained on=AffectNet-82023.12 | 0.9117 | |
| S2Dyear=2024, Pre-trained=AffectNet-82024.11 | 0.9117 | |
| PF-ViT2023.12 | 0.9116 | |
| PF-VITbackbone=ViT-B, pre-training data=AffectNet-270K, venue=TCSS'242022.07 | 0.9116 | |
| FG-AGRbackbone=Hybrid ViT, pre-training data=MS-Celeb-1M, venue=TCSVT'232022.07 | 0.9109 | |
| S2D w/o TMA2023.12 | 0.9101 | |
| TAN2023.12 | 0.91 | |
| TANyear=20232024.11 | 0.91 | |
| APViTTransformer-based=true2022.12 | 0.9086 | |
| APViT# Params=65.2M2023.06 | 0.9086 | |
| APViT2023.12 | 0.9086 | |
| APVITyear=20222024.11 | 0.9086 | |
| TransFERTransformer-based=true2022.12 | 0.9083 | |
| TransFER# Params=65.2M2023.06 | 0.9083 | |
| TransFER2023.12 | 0.9083 | |
| TransFERbackbone=Hybrid ViT, pre-training data=MS-Celeb-1M, venue=ICCV'212022.07 | 0.9083 | |
| TransFERyear=20212024.11 | 0.9083 | |
| DDAMFN# Params=4.11M2023.06 | 0.9074 | |
| KTNTransformer-based=false2022.12 | 0.9049 | |
| MTACbackbone=Swin-S [58], pre-training data=ImageNet1K, venue=TMM'242022.07 | 0.904 | |
| MAE-VITbackbone=ViT-B, pre-training data=AffectNet-270K, venue=TCSS'242022.07 | 0.9018 | |
| EAC# Params=11.2M2023.06 | 0.8964 | |
| EAC2023.12 | 0.8964 | |
| EACbackbone=ResNet-18, pre-training data=MS-Celeb-1M, venue=ECCV'222022.07 | 0.8964 | |
| EACyear=20222024.11 | 0.8964 | |
| MRANbackbone=Hybrid ViT, pre-training data=MS-Celeb-1M, venue=TCSVT'232022.07 | 0.8959 | |
| DMUEBackbone=ResNet-50IBN2021.04 | 0.8951 | |
| DMUEBackbone=ResNet-50IBN2021.04 | 0.8951 | |
| DMUE2023.12 | 0.8951 | |
| DMUEbackbone=Res-50IBN, pre-training data=Ms-Celeb-1M, venue=CVPR'212022.07 | 0.8951 | |
| DMUEyear=20212024.11 | 0.8951 | |
| SCAN-CCI# Params=70M2023.06 | 0.8942 | |
| SCNTransformer-based=false2022.12 | 0.8935 | |
| SCN2023.12 | 0.8935 | |
| SCNbackbone=ResNet-18, pre-training data=Ms-Celeb-1M, venue=CVPR'202022.07 | 0.8935 | |
| MAE-VITbackbone=ViT-S, pre-training data=AffectNet-270K, venue=TCSS'242022.07 | 0.8935 | |
| MViTTransformer-based=true2022.12 | 0.8922 | |
| MVT2023.12 | 0.8922 | |
| MVTbackbone=DeiT-S [44], pre-training data=ImageNet1K, venue=Arxiv'212022.07 | 0.8922 | |
| RAN-VGG16Network=VGG16, Pre-trained Dataset=VGG_Face, Label distribution supervision=true2019.05 | 0.8916 | |
| RAN-VGG16Transformer-based=false2022.12 | 0.8916 | |
| RAN# Params=11.2M2023.06 | 0.8916 | |
| VTFFTransformer-based=true2022.12 | 0.8881 | |
| VTFF# Params=80.1M2023.06 | 0.8881 | |
| VTFF2023.12 | 0.8881 | |
| VTFFbackbone=Hybrid ViT, pre-training data=MS-Celeb-1M, venue=TAFFC'232022.07 | 0.8881 | |
| [1]Network=SeNet50, Pre-trained Dataset=VGG_Face2, Label distribution supervision=true2019.05 | 0.888 | |
| SeNet50Data Annotation Type=extra distribution data2021.04 | 0.888 | |
| SeNet50Backbone=SeNet50, Extra distribution data=true2021.04 | 0.888 | |
| SeNet50Transformer-based=false2022.12 | 0.888 | |
| MAE-VITbackbone=ViT-T, pre-training data=AffectNet-270K, venue=TCSS'242022.07 | 0.8867 | |
| DMUEBackbone=ResNet-182021.04 | 0.8864 | |
| DMUEBackbone=ResNet-182021.04 | 0.8864 | |
| DMUEbackbone=ResNet-18, pre-training data=Ms-Celeb-1M, venue=CVPR'212022.07 | 0.8864 | |
| RAN-ResNet18Network=ResNet18, Pre-trained Dataset=MS_Celeb_1M2019.05 | 0.8855 | |
| RAN2021.04 | 0.8855 | |
| RAN2021.04 | 0.8855 | |
| RANTransformer-based=false2022.12 | 0.8855 | |
| IPD-FERbackbone=ResNet-18, pre-training data=CASIA-WebFace [53], venue=TAFFC'222022.07 | 0.8842 | |
| SCN2021.04 | 0.8801 | |
| SCN2021.04 | 0.8801 | |
| SCNyear=20202024.11 | 0.8801 | |
| CNNs and BOVW + local SVMData Augmentation (aug.)=false2018.04 | 0.8776 | |
| [27]Network=ResNet18+VGG16, Pre-trained Dataset=None2019.05 | 0.874 | |
| Res+VGG2021.04 | 0.874 | |
| Res+VGGBackbone=ResNet + VGG2021.04 | 0.874 | |
| CNNs and BOVW + local SVMData Augmentation (aug.)=true2018.04 | 0.8725 | |
| ESR-9Ensemble size=92020.01 | 0.8715 | |
| EACVenue=ECCV'222024.10 | 0.8703 | |
| CNNs and BOVW + global SVMData Augmentation (aug.)=true2018.04 | 0.8696 | |
| CNNs and BOVW + global SVMData Augmentation (aug.)=false2018.04 | 0.8668 | |
| PTTransformer-based=false2022.12 | 0.866 | |
| SHCNN2020.01 | 0.8654 | |
| SCNVenue=CVPR'202024.10 | 0.8597 | |
| [5]Network=VGG13, Pre-trained Dataset=None, Label distribution supervision=true2019.05 | 0.851 | |
| PLDData Annotation Type=extra distribution data2021.04 | 0.851 | |
| PLDExtra distribution data=true2021.04 | 0.851 | |
| PLDTransformer-based=false2022.12 | 0.851 | |
| VGG16-PLDLoss/Label=Probabilistic Label Distribution2020.01 | 0.8499 | |
| Barsoum et al.Data Augmentation (aug.)=true2018.04 | 0.8499 | |
| VGG16-CELLoss/Label=Cross-Entropy Loss2020.01 | 0.8472 | |
| TFE-JL2020.01 | 0.843 | |
| TFE-joint learningTransformer-based=false2022.12 | 0.8429 | |
| VGG16-MLAggregator=ML2020.01 | 0.8397 | |
| VGG16-MVAggregator=Majority Voting2020.01 | 0.8385 | |
| RANVenue=TIP'202024.10 | 0.8363 | |
| ResNet18 + FCStructure=ResNet18 with Fully Connected layer2020.01 | 0.834 | |
| ResNet18Backbone=ResNet-182020.01 | 0.831 |