What are the advantages of deep learning for face recognition compared with traditional methods?

In traditional methods, feature extraction mainly depends on the artificially designed extractor, which requires professional knowledge and complicated parameter adjustment process. At the same time, each method is aimed at a specific application, and its generalization ability and robustness are poor.

Deep learning is mainly data-driven feature extraction. According to a large number of samples, we can get in-depth and data-specific feature representation, which is more efficient and accurate for the expression of data sets, and the extracted abstract features are more robust and have better generalization ability, which can be end-to-end. The disadvantage is that the sample set has great influence and requires high computing power.