Speaker Identification in a Crowd on LRS3 (test)
87AccuracyGestSync
Evaluation Results
| Method | Links | |
|---|---|---|
| GestSyncInput representation=RGB, Number of negative speakers=1-negative2023.10 | 87 | |
| GestSyncInput representation=Keypoint-image, Number of negative speakers=1-negative2023.10 | 82.4 | |
| GestSyncInput representation=Keypoint-vector, Number of negative speakers=1-negative2023.10 | 79 | |
| GestSyncInput representation=RGB, Number of negative speakers=3-negatives2023.10 | 77.6 | |
| GestSyncInput representation=RGB, Number of negative speakers=5-negatives2023.10 | 73.2 | |
| GestSyncInput representation=Keypoint-image, Number of negative speakers=3-negatives2023.10 | 70.6 | |
| GestSyncInput representation=Keypoint-image, Number of negative speakers=5-negatives2023.10 | 62.6 | |
| GestSyncInput representation=Keypoint-vector, Number of negative speakers=3-negatives2023.10 | 58.7 | |
| GestSyncInput representation=Keypoint-vector, Number of negative speakers=5-negatives2023.10 | 50.1 |