Video Action Recognition on Breakfast
87.5Top-1 AccuracyMoTIF
Evaluation results
| Method | Links | |
|---|---|---|
| MoTIFBackbone=PE-G/142025.09 | 87.5 | |
| MoTIF-STBackbone=PE-G/142025.09 | 87.3 | |
| MoTIFBackbone=PE-L/142025.09 | 83.2 | |
| Ours-cumulativeBackbone=ResNet50, #Frames=all, #Clusters=162021.03 | 76.6 | |
| Global CBMBackbone=PE-G/14, Evaluation Protocol=Supervised2025.09 | 75.8 | |
| Ours-slopeBackbone=ResNet50, #Frames=all, #Clusters=162021.03 | 74.9 | |
| TSM (our impl.)Backbone=ResNet50, #Frames=162021.03 | 72.1 | |
| 3D ResNet-50 + TimeceptionBackbone=3D ResNet-50, #3D=true, #Frames=5122021.03 | 71.3 | |
| MoTIFBackbone=ViT-L/142025.09 | 71 | |
| I3D + VideoGraphBackbone=3D Inception-v1, #3D=true, #Frames=5122021.03 | 69.5 | |
| I3D + ActionVLADBackbone=3D Inception-v1, #3D=true, #Frames=5122021.03 | 65.5 | |
| NoFrame2025.09 | 62 | |
| VideoGraphBackbone=ResNet152, #Frames=642021.03 | 59.1 | |
| TSM2025.09 | 59.1 | |
| I3DBackbone=3D Inception-v1, #3D=true, #Frames=5122021.03 | 58.6 | |
| ActionVLADBackbone=ResNet152, #Frames=642021.03 | 55.5 | |
| Zero-shotBackbone=PE-G/14, Evaluation Protocol=Zero-shot2025.09 | 47.4 | |
| ResNet152Backbone=ResNet152, #Frames=642021.03 | 41.1 |