Table 3 Performance for video classification only.
From: Understanding action concepts from videos and brain activity through subjects’ consensus
Dense trajectories: HOG | 18.64% | ResNet-50 | 29.33% |
Dense trajectories: HOF | 16.95% | Temporal residual networks (\(N=4\)) | 28.30% |
Dense trajectories: MBHx | 23.33% | Temporal residual networks (\(N=8\)) | 31.70% |
Dense trajectories: MBHy | 26.67% | Temporal shift models | 42.33% |
