Self-taught learning of a deep invariant representation for visual tracking via temporal slowness principle

Visual representation is crucial for visual tracking method s performances. Conventionally, visual representations adopted in visual tracking rely on hand-crafted computer vision descriptors. These...