Face Detection and Recognition from Video using Deep Learning

Facial recognition (FR) and verification are currently the most effective techniques for detecting illegal activity. It can be used in a wide range of applications from criminal identity, security and surveillance to entertainment websites. For verifying consumers, this method (facial recognition) is very useful in banks, airports and other organizations. Convolutional neural networks (CNNs) have sparked interest in deep learning for face recognition, but training CNNs requires more data and applications such as criminal activity (robbery, murder, etc.). It’s a big problem when it comes to relationships. Therefore, this study provided a facial recognition system that effectively supports law enforcement and administration by making criminal investigations easier, faster, and more time-consuming. In this study, face recognition from videos is performed using a pre-trained model called FaceNet (FN). Face images are transformed by FN into a compact Euclidean space with extended inter-face distances. We are attempting to create a medisale app utilizing this concept, in which we register medical store owners by doing KYC verification on them. The KYC procedure uses real-time video and still images of the user to perform face landmark detection. This entire procedure takes ten seconds, but we can make it faster.