Disguised Facial Recognition Using Neural Networks

In this paper, we present a real-time deep neural network architecture (called DiFRuNNT) for disguised face verification. The proposed model consists of two neural networks, first one being a convolutional neural network (CNN) that predicts 20 facial key-points in the image and the second neural network classifies the subject based on the angles and ratios calculated from the predicted points. The accuracies are 67.4% and 74.8% for prediction and classification respectively and the results have been compared with the state-of-the-art methodologies also.