Anti-Spoofing System for Face Detection Using Convolutional Neural Network

The concept of face anti-spoofing is an important part in the face recognition system. It has great importance for fiscal payment and different networking systems in today’s modern world. A new system has been introduced using a three-layer convolutional neural network. Accordingly, in this paper, we present a deep neural network strategy for face anti-spoofing. This paper proposes a system of detecting spoofing using convolutional neural network (CNN) classifier. The convolutional neural network system is constructed to arrest the spoofed faces from piercing in the name of genuine person. We have considered 3 layers of CNN in order to make the detection of the images in a more clear format. Self-datasets of real and fake images are created to train the neural network. The two datasets are trained singly to resolve the absolute outgrowth. The accuracy achieved by our model is quite satisfactory. The experimental results over the validation dataset and training dataset show that this system shows better performance and has demonstrated a satisfactory accuracy over other models.