Action Unit Detection on Aff-Wild2 (val)
58F1-score PAUCLIP+TCN
Evaluation results
| Method | Links | |
|---|---|---|
| CLIP+TCN2026.03 | 58 | |
| CLIP+TCN2026.03 | 58 | |
| MAE ViT + TCN2026.03 | 57.6 | |
| MAE ViT + TCN2026.03 | 57.6 | |
| MAE+Transformer Fusion2026.03 | 56.9 | |
| MAE+Transformer Fusion2026.03 | 56.9 | |
| ConvNeXt + Whisper+STN +TCN + Sliding Window2026.03 | 56.1 | |
| ConvNeXt + Whisper+STN +TCN + Sliding Window2026.03 | 56.1 | |
| EmotiEffNetInput Features=logits + embeddings, Smoothing=true, Thresholding Strategy=best thresholds2026.03 | 54.7 | |
| EmotiEffNetInput=logits + embeddings, Smoothing=true, Threshold=best thresholds2026.03 | 54.7 | |
| EmotiEffNetInput Features=logits + embeddings, Smoothing=true, Thresholding Strategy=0.52026.03 | 54.6 | |
| EmotiEffNetInput=logits + embeddings, Smoothing=true, Threshold=0.52026.03 | 54.6 | |
| Best EfficientNet2026.03 | 54.5 | |
| Best EfficientNet2026.03 | 54.5 | |
| EmotiEffNetInput Features=embeddings, Smoothing=true, Thresholding Strategy=best thresholds2026.03 | 54.2 | |
| EmotiEffNetInput=embeddings, Smoothing=true, Threshold=best thresholds2026.03 | 54.2 | |
| EmotiEffNetInput Features=embeddings, Smoothing=true, Thresholding Strategy=0.52026.03 | 54 | |
| EmotiEffNetInput=embeddings, Smoothing=true, Threshold=0.52026.03 | 54 | |
| EmotiEffNetInput Features=logits + embeddings, Smoothing=false, Thresholding Strategy=best thresholds2026.03 | 53.9 | |
| EmotiEffNetInput=logits + embeddings, Threshold=best thresholds2026.03 | 53.9 | |
| AUD-TGN2026.03 | 53.7 | |
| EmotiEffNetInput Features=logits + embeddings, Smoothing=false, Thresholding Strategy=0.52026.03 | 53.7 | |
| AUD-TGN2026.03 | 53.7 | |
| EmotiEffNetInput=logits + embeddings, Threshold=0.52026.03 | 53.7 | |
| EmotiEffNetInput Features=embeddings, Smoothing=false, Thresholding Strategy=best thresholds2026.03 | 52.9 | |
| EmotiEffNetInput=embeddings, Threshold=best thresholds2026.03 | 52.9 | |
| EmotiEffNetInput Features=embeddings, Smoothing=false, Thresholding Strategy=0.52026.03 | 52.7 | |
| EmotiEffNetInput=embeddings, Threshold=0.52026.03 | 52.7 | |
| EmotiEffNetInput Features=logits, Smoothing=true, Thresholding Strategy=best thresholds2026.03 | 49.3 | |
| EmotiEffNetInput=logits, Smoothing=true, Threshold=best thresholds2026.03 | 49.3 | |
| EmotiEffNetInput Features=logits, Smoothing=true, Thresholding Strategy=0.52026.03 | 48.6 | |
| EmotiEffNetInput=logits, Smoothing=true, Threshold=0.52026.03 | 48.6 | |
| EmotiEffNetInput Features=logits, Smoothing=false, Thresholding Strategy=best thresholds2026.03 | 48.3 | |
| EmotiEffNetInput=logits, Threshold=best thresholds2026.03 | 48.3 | |
| EmotiEffNetInput Features=logits, Smoothing=false, Thresholding Strategy=0.52026.03 | 47.5 | |
| EmotiEffNetInput=logits, Threshold=0.52026.03 | 47.5 | |
| DDAMFN+LSTMThresholding Strategy=best thresholds2026.03 | 45.1 | |
| DDAMFN+LSTMThreshold=best thresholds2026.03 | 45.1 | |
| CLIP2026.03 | 43 | |
| CLIP2026.03 | 43 | |
| DDAMFN+LSTMThresholding Strategy=0.52026.03 | 41.1 | |
| DDAMFN+LSTMThreshold=0.52026.03 | 41.1 | |
| Baseline VGGFACE2026.03 | 39 | |
| Baseline VGGFACE2026.03 | 39 | |
| wav2vec 2.02026.03 | 31.3 | |
| wav2vec 2.02026.03 | 31.3 |