Face Detection and Recognition From Distance Based on Deep Learning

Abstract

Face recognition is an important biometric in video surveillance. However, the conventional algorithms of face recognition are also susceptible to various conditions. The contribution of this chapter is to explore face recognition by using deep learning, including detecting the location of human faces on the given images at various distances and multiple angles. In the distances, the influence of the distance from the camera to a face and the size of the face in the images is explored. There are 500 images collected from various videos as the input and they are applied to train the proposed models. The accuracy of face recognition from the videos excluding training dataset is at 90.18%. The results indicate that this method is able to recognise human faces with partial occlusion and various distances.