Face recognition in video using Dempster-Shafer theory
Samuel Foucher, L. Gagnon · 2004
We propose a new approach based on the Dempster-Shafer theory to increase identification performance of character faces in real films. We use a frame-based approach where faces are detected and identified independently in each frame without taking into account any temporal information. Statistical evidence about character faces is accumulated within the shot. Tests on documentary films show promising results both in terms of high detection and low false alarm rates.