Neural Network Projects with Python

Future work

As we saw, our face recognition system certainly works well under simple conditions. However, it is definitely not fool-proof, and certainly not secure enough to be implemented in important applications. For one, the face detection system can be fooled by a static photo (try it yourself!). Theoretically, that means we can bypass the authentication by placing a static photo of an authorized user in front of the webcam.

Techniques to solve this problem are known as anti-spoofing techniques. Anti-spoofing techniques are a keenly studied area in face recognition. In general, there are two main anti-spoofing techniques used today:

• Liveness detection: Since a photo is a static two-dimensional image and a real face is dynamic and three-dimensional, ...