Automatic Face Recognition For Home Safety Using Video-Based Side-View Face Images

What is it about?

Face recognition from side-view positions is an essential task for recognition systems with real-world scenarios. In this paper, we first contribute a newly collected face database, named UT-DOOR, where we recorded 98 subjects with four cameras attached at doorposts as they pass through doors. Secondly, we propose a face recognition system for home safety applications, where we automatically detect and recognize faces using side-view images in videos.

Featured Image

Why is it important?

Due to the complex structure of the human face, face recognition up to side-view is a challenging problem. To tackle this problem, we first collected a large-scale video face database, named UT-DOOR. Compared to other available face databases, UT-DOOR consists of video recordings with a significant number of side-view images. Secondly, we propose an automatic side-view face recognition system for home safety applications. Such a system can help to increase the situational awareness and prevent factors that may cause further accidents or detect an emergency in time. Experimental results show that our system can successfully recognize side-view faces from videos.