Real-Time Audio-Visual Analysis for  Multiperson Videoconferencing

<table>Recall versus precision for face detection, face tracking, and speaker match (Dataset 2). Speaker match shows lower performance in case of Dataset 2 due to the presence of 4 participants within a sector of 100°.</table>

Advances in Multimedia

fig13

Figure 13

Figure 13: Real-Time Audio-Visual Analysis for  Multiperson Videoconferencing 