Improving face recognition from a single image per person via virtual images produced by a bidirectional network

In this article, for the purpose of improving neural network models applied in face recognition using single image per person, a bidirectional neural network inspired of neocortex functional model is presented. We have applied this novel adapting model to separate person and pose information. To increase the number of training samples in the classifier neural network, virtual views of frontal images in the test dataset are synthesized using estimated manifolds. Training classifier network via virtual images gives an accuracy rate of 85.45% which shows 14.55% improvement in accuracy of face recognition compared to training classifier with only frontal view images.