Egocentric Activity Recognition on GTEA Gaze+ (leave-one-subject-out cross val)
68.67Accuracy*Gaze-combined Attention
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Gaze-combined AttentionBackbone network=I3D2020.11 | 68.67 | 64.81 | |
| Gaze-combined AttentionBackbone network=I3D [3]2020.11 | 68.67 | 64.81 | |
| Shen et al.Backbone network=SSD+LSTM2020.11 | 67.1 | — | |
| Shen et al.Backbone network=SSD+LSTM [19]2020.11 | 67.1 | — | |
| Ma et al.Backbone network=FCN32s+CNN-M-20482020.11 | 66.4 | — | |
| Ma et al.Backbone network=FCN32s+CNN-M-2048 [20, 5]2020.11 | 66.4 | — | |
| Sudhakaran et al.Backbone network=ResNet34+LSTM2020.11 | 60.13 | — | |
| Sudhakaran et al.Backbone network=ResNet34+LSTM [10, 36]2020.11 | 60.13 | — | |
| MCNBackbone network=I3D2020.11 | — | 61.14 | |
| MCNBackbone network=I3D [3]2020.11 | — | 61.14 |