Facial Expression Recognition on FER 2013 (test)
79.79Accuracy RateResEmoteNet
Evaluation results
| Method | Links | ||
|---|---|---|---|
| ResEmoteNet2024.09 | 79.79 | — | |
| En. ResMaskingNet2024.09 | 76.82 | — | |
| EmoNeXt2024.09 | 76.12 | — | |
| EmoNeXtSize=XLarge2025.01 | 76.12 | — | |
| Seg. VGG-192024.09 | 75.97 | — | |
| Segmentation VGG-19Backbone=VGG-192025.01 | 75.97 | — | |
| EmoNeXtSize=Large2025.01 | 75.57 | — | |
| CNNs and BOVW + local SVMData Augmentation (aug.)=true2018.04 | 75.42 | — | |
| CNNS + BOVW2025.01 | 75.42 | — | |
| CNNs and BOVW + local SVMData Augmentation (aug.)=false2018.04 | 74.92 | — | |
| EmoNeXtSize=Base2025.01 | 74.91 | — | |
| LHC-Net2025.01 | 74.42 | — | |
| EmoNeXtSize=Small2025.01 | 74.33 | — | |
| LHC-NetC2025.01 | 74.28 | — | |
| ConvNeXtSize=XLarge2025.01 | 74.15 | — | |
| Residual Masking Network2025.01 | 74.14 | — | |
| ConvNeXtSize=Large2025.01 | 73.46 | — | |
| Connie et al.Data Augmentation (aug.)=true2018.04 | 73.4 | — | |
| CNNs and BOVW + global SVMData Augmentation (aug.)=false2018.04 | 73.34 | — | |
| EmoNeXtSize=Tiny2025.01 | 73.34 | — | |
| VGG2025.01 | 73.28 | — | |
| CNNs and BOVW + global SVMData Augmentation (aug.)=true2018.04 | 73.25 | — | |
| ConvNeXtSize=Base2025.01 | 73.22 | — | |
| ResNet50Backbone=ResNet-502025.01 | 73.2 | — | |
| Kim et al.Data Augmentation (aug.)=true2018.04 | 72.72 | — | |
| SE-Net50Backbone=ResNet-502025.01 | 72.5 | — | |
| ConvNeXtSize=Small2025.01 | 72.34 | — | |
| Ad-Corre2025.01 | 72.03 | — | |
| Yu et al.Data Augmentation (aug.)=true2018.04 | 72 | — | |
| ConvNeXtSize=Tiny2025.01 | 71.99 | — | |
| Hua et al.Data Augmentation (aug.)=true2018.04 | 71.91 | — | |
| Inception2025.01 | 71.6 | — | |
| TangData Augmentation (aug.)=true2018.04 | 71.16 | — | |
| Li et al.Data Augmentation (aug.)=true2018.04 | 70.66 | — | |
| Deep-Emotion2019.02 | 70.02 | — | |
| Deep Emotion2025.01 | 70.02 | — | |
| ALTEvaluation protocol=Fully supervised2022.11 | 69.85 | — | |
| EfficientNetB2Params (M)=9.22026.01 | 68.78 | — | |
| EfficientNetB2-based FER modelNumber of parameters=~9.2M2026.01 | 68.78 | — | |
| FSNEvaluation protocol=Fully supervised2022.11 | 67.6 | — | |
| Ionescu et al.Data Augmentation (aug.)=false2018.04 | 67.48 | — | |
| Bag of Words2019.02 | 67.4 | — | |
| VGG16 (trained from scratch)Params (M)=1382026.01 | 67.23 | — | |
| Mollahosseini et al2019.02 | 66.4 | — | |
| VGG+SVM2019.02 | 66.31 | — | |
| GoogleNet2019.02 | 65.2 | — | |
| GoogleNet2025.01 | 65.2 | — | |
| AGRABackbone=ResNet502025.12 | 58.95 | — | |
| EfficientNet (prior reports)Params (M)=4–202026.01 | 58 | — | |
| PCL (Ours)Evaluation protocol=Self-supervised (linear evaluation)2022.11 | 56.81 | — | |
| ECANBackbone=ResNet502025.12 | 56.46 | — | |
| CSRLBackbone=ResNet182025.12 | 55.53 | — | |
| MotivNetBackbone=ViT2025.12 | 53.87 | — | |
| SimCLREvaluation protocol=Self-supervised (linear evaluation), reproduced by authors=true2022.11 | 49.51 | — | |
| FaceCycleEvaluation protocol=Self-supervised (linear evaluation)2022.11 | 48.76 | — | |
| BMVC'20Evaluation protocol=Self-supervised (linear evaluation)2022.11 | 47.61 | — | |
| MoCoEvaluation protocol=Self-supervised (linear evaluation)2022.11 | 47.24 | — | |
| FAb-NetEvaluation protocol=Self-supervised (linear evaluation)2022.11 | 46.98 | — | |
| HoGEvaluation protocol=Self-supervised (linear evaluation)2022.11 | 45.47 | — | |
| TCAEEvaluation protocol=Self-supervised (linear evaluation)2022.11 | 45.05 | — | |
| LBPEvaluation protocol=Self-supervised (linear evaluation)2022.11 | 37.89 | — |