Facial Expression Recognition On Acted Facial
Metrics
Accuracy(on validation set)
Results
Performance results of various models on this benchmark
| Paper Title | ||
|---|---|---|
| LResNet50E-IR (5 models with augmentation) | 65.5% | Exploring Emotion Features and Fusion Strategies for Audio-Video Emotion Recognition |
| ResNet50 | 65.5% | Exploring Emotion Features and Fusion Strategies for Audio-Video Emotion Recognition |
| EAC | 65.32% | Learn From All: Erasing Attention Consistency for Noisy Label Facial Expression Recognition |
| LResNet50E-IR (1 model with augmentation) | 63.7% | Exploring Emotion Features and Fusion Strategies for Audio-Video Emotion Recognition |
| LResNet50E-IR (1 model) | 61.1% | Exploring Emotion Features and Fusion Strategies for Audio-Video Emotion Recognition |
| Multi-task EfficientNet-B0 | 59.27 | Facial expression and attributes recognition based on multi-task learning of lightweight neural networks |
| resnet18_noisy | 55.17% | Noisy Student Training using Body Language Dataset Improves Facial Expression Recognition |
| resnet18 | 51.181% | Frame attention networks for facial expression recognition in videos |
0 of 8 row(s) selected.