A Virtual Environment Tool for Benchmarking Face Analysis Systems

Abstract

In this article, a virtual environment for realistic testing of face analysis systems under uncontrolled conditions is proposed. The key elements of this tool are a simulator, and real face and background images taken under real-world conditions with different acquisition conditions, such as indoor or outdoor illumination. Inside the virtual environment, an observing agent, the one with the ability to recognize and detect faces, can navigate and observe the face images, at different distances, and angles. During the face analysis process, the agent can actively change its viewpoint and relative distance to the faces in order to improve the recognition results. The virtual environment provides all behaviors to the agent (navigation, positioning, face’s image composing under different angles, etc.), except the ones related with the analysis of faces (detection, recognition, pose estimation, etc.). In addition we describe different kinds of experiments that can be implemented for quantifying the face analysis capabilities of agents and provide usage example of the proposed tool in evaluating a face recognition system in a service robot.