Automated Face Spoofing Detection using Machine Learning: A Review

Face spoofing/ liveliness is using someone's identity for doing criminal activities. Taking a survey from past years, face spoof attacks have become very dangerous to society specially in bank frauds, social media etc. Many of the techniques have been used in past years to overcome these attacks, but the drawbacks are difficult to remove in these methods. These face spoof attacks include printing attacks, video attacks, 3-dimensional mask attacks. In printing attacks, the attacker uses the printed picture of the victim so as to look like him on the camera. In the video attack, the attacker uses a short video clip of the victim that is a little bit tougher than printing attack. In 3D mask attack, the attacker uses a mask on his face that is made with the resin materials that are highly destructive and these masks have a very sleek surface. In this paper, we are using machine learning, Convolutional Neural Network with tensor flow to detect the spoofing attacks using neural networks which is having higher accuracy as compared to other techniques.