| Summary: | Robust visual tracking plays an important role in many applications such as security surveillance, human-computer interaction and video analytics. Given the position of a target in the first frame of a video clip, the objective is to track the target in following frames of this sequence. Although many promising trackers have been proposed and achieved fairly good performance in simple environment, it is still very challenging to efficiently track arbitrary objects in complicated situations, especially when appearance changes significantly and heavy occlusion occurs. In this thesis we present four different tracking algorithms which exploit the sparse coding, part-based model, color feature learning and convolutional network features to handle the aforementioned challenges.Extensive experiments have been done respectively to prove the effectiveness of our proposed trackers.
|
|---|