Real-Time Audio-Visual Analysis for Multiperson Videoconferencing

Figure 12

Recall versus precision for face detection, face tracking, and speaker match (Dataset 1). Both face tracking and speaker match show good performance as there are only two participants within a sector of 100°.