In recent years, video surveillance has become increasingly dependent on automated human tracking via camera networks. Not only is it difficult to follow people over camera networks since people's appearances change over time, but there are also a lot of real-world uses for this technology, including security surveillance, retail, and medical. Due to differences in appearance brought on by changes in viewpoint and illumination, background, occlusion, non-rigid deformations, and intra-class heterogeneity in shape and position, human tracking is one of the most difficult components of the task.