Method | CMU | SFU |
---|
 | Frame | Sequence | Frame | Sequence |
---|
K-means [6] | – | 0.133 | – | 0.474 |
Bayesian GMM [6] | – | 0.133 | – | 0.231 |
One-class SVM [25] | – | 0.400 | – | 0.356 |
Bayesian GMM [25] | – | 0.267 | – | 0.350 |
Ours (leaky ReLU) | 0.233 |
0.067
| 0.253 |
0.158
|
- The two methods [6, 25] are not adaptive to perform per-frame assessment
- The italic values indicate the best results