Facial Expression Recognition on CK+ (10-fold cross-val)
99.8AccuracyViT + SE
Evaluation Results
| Method | Links | |
|---|---|---|
| ViT + SEModel Type=Deep Learning2021.07 | 99.8 | |
| [38]Model Type=Deep Learning, Year=20212021.07 | 98 | |
| [9]Model Type=Deep Learning, Year=20202021.07 | 97.59 | |
| PPDNEvaluation Protocol=10-fold cross-validation, Input Setting=Nearly peak expressions (last 3 frames)2016.07 | 97.3 | |
| BDBNEvaluation Protocol=10-fold cross-validation, Input Setting=Nearly peak expressions (last 1-3 frames)2016.07 | 96.7 | |
| LBPSVMEvaluation Protocol=10-fold cross-validation, Input Setting=Nearly peak expressions (last 1-3 frames)2016.07 | 95.1 | |
| [2]Model Type=Handcrafted, Year=20142021.07 | 95.03 | |
| GoogLeNet (baseline)Evaluation Protocol=10-fold cross-validation, Optimizer=SGD2016.07 | 95 | |
| AdaGaborEvaluation Protocol=10-fold cross-validation, Input Setting=Nearly peak expressions (last 1-3 frames)2016.07 | 93.3 | |
| CSPLEvaluation Protocol=10-fold cross-validation, Input Setting=Nearly peak expressions (last 1-3 frames)2016.07 | 89.9 |