Comparison of adaptive appearance methods for tracking faces in video surveillance

Face recognition is increasingly employed by public safety organizations in decision support systems for video surveillance, to detect the presence of individuals of interest. In the context of spatiotemporal face recognition, tracking is an important function used to locate, follow and regroup faces of different individuals in a scene. Techniques for face tracking in video surveillance should be robust to changes in pose, expression and illumination, as well as occlusion in cluttered scenes. Given these challenges, trackers based on adaptive appearance modelling (AAM) typically improve target's state estimation because they initiate and update an internal face model per individual according to changes in facial appearance. In this paper, the performance of three AAM trackers - Incremental Visual Tracking (IVT), Tracking Learning Detection (TLD) and Discriminative Sparse Coding based Tracking (DSCT) - are compared for face tracking with video surveillance applications in mind. These methods are evaluated according to area overlap error, tracking error and time complexity using Chokepoint videos collected in uncontrolled video-surveillance environments, where individuals walk through portals. Results indicate that IVT outperforms the others in its ability to accurately track faces in the presence of occlusion, and under variations in pose, scale and lighting. Further characterization of IVT indicates that using a small batch size and forgetting factor during update provide better tracking accuracy when face tracks changes in their capture conditions. When conditions change more gradually, IVT benefits from assessing facial quality before updating face models.