机器人学研究室知识库

Multi-task and low rank learning methods have attracted increasing attention for visual tracking. However, most trackers only focus on learning appearance subspace basis or the sparse low rankness of representation, thus do not make full use of the structure information among and inside target candidates (or samples). In this work, we propose a dual graph regularized discriminative low rank learning for multi-task tracker, which integrates the discriminative subspace and intrinsic geometric structures among tasks. By constructing double graphs regula- tions from two views of multi-task observation, the developed modal not only exploits the intrinsic relationship among tasks, and preserves the spatial layout structure among the local patches inside each candidate, but also learns the salient features of the target samples. This operation is benefit to having good target representation and improving the performance of the tracker. Moreover, our developed tracker is a collaborate multi- task tracking model, and learns the discriminative subspace with adaptive dimension and optimal classifier simultaneously. Then, a collaborate metric is developed to find the best candidate, which integrates both classification reliability and representation accu- racy. Encouraging experimental results on a large set of public video sequences justify that our tracker performs favourably against many other state-of-the-art trackers.