Facial Expression Recognition on AffectNet (val)
61.6AccuracyAVER
Evaluation Results
| Method | Links | |
|---|---|---|
| AVERModel Type=Student model, Knowledge Distillation=true, Unlabeled Data Usage (PowderFaces)=true2021.03 | 61.6 | |
| AVERModel Type=Teacher model2021.03 | 61.3 | |
| AVERModel Type=Student model, Knowledge Distillation=true, Unlabeled Data Usage (PowderFaces)=false2021.03 | 61.1 | |
| Georgescu et al.2021.03 | 59.6 | |
| Siqueira et al.2021.03 | 59.3 | |
| AVERModel Type=Student model, Knowledge Distillation=false2021.03 | 58.8 | |
| FIELDSEvaluation protocol=KNN, k (number of neighbors)=5, Distance weighting=distance weighting, Input features=expression and jaw parameters2025.11 | 51.4 | |
| EMOCAEvaluation protocol=KNN, k (number of neighbors)=5, Distance weighting=distance weighting, Input features=expression and jaw parameters2025.11 | 47.2 | |
| SMIRKEvaluation protocol=KNN, k (number of neighbors)=5, Distance weighting=distance weighting, Input features=expression and jaw parameters2025.11 | 44.8 | |
| TEASEREvaluation protocol=KNN, k (number of neighbors)=5, Distance weighting=distance weighting, Input features=expression and jaw parameters2025.11 | 42.9 |