Research Article
Design and Analysis of an Expert System for the Detection and Recognition of Criminal Faces
Table 3
Comparison of facial methods.
| ā | LBPH | SVM | CNN |
| Advantages | It recognizes both the frontal and side view of faces | It works well with high-dimensional data | It automatically detects features without human supervision | Disadvantage | With large-scale databases, the histograms produced are too long and increase computation time | When using a large database, the training time can be too much. The overlapping classes can decrease the accuracy | Accuracy decreases when images have different poses, backgrounds, and lighting |
|
|