TABLE 4.
Classification performance of the proposed L-WLMP algorithm for video and frame based presentation attack detection on the proposed Snapchat database. The results are reported in terms of the equal error rate and average classification accuracy (%).
| Input | BSIF filter bit | EER | ACER |
|---|---|---|---|
| Video | 5 | 19.9 | 22.3 |
| 6 | 21.8 | 23.2 | |
| 7 | 18.7 | 21.9 | |
| 8 | 18.2 | 20.6 | |
| 7 and 8 | 17.6 | 19.5 | |
| 5, 7, and 8 | 17.9 | 19.6 | |
| 5, 6, 7, and 8 | 18.0 | 20.0 | |
| 5, 6, 7, and 8 + PCA | 23.2 | 24.5 | |
| Frame | 5 | 26.2 | 27.3 |
| 6 | 27.3 | 28.1 | |
| 7 | 25.7 | 27.2 | |
| 8 | 24.9 | 26.3 | |
| 7 and 8 | 24.6 | 25.9 | |
| 5, 7, and 8 | 24.7 | 26.0 | |
| 5, 6, 7, and 8 | 24.7 | 26.0 | |
| 5, 6, 7, and 8 + PCA | 28.7 | 29.6 |
Result of the best performing algorithm is highlighted in bold.