Metric learning with trace-norm regularization for person re-identification

Person re-identification is a challenging problem in multicamera surveillance system. Existing methods always make use of metric learning to model the appearance variations of pedestrians between different cameras. However, these methods ignore the complexity of appearance variations in PRID and the learned models are liable to be over-fitted. In this paper, we propose large margin nearest neighbor with trace-norm regularization (LMNN-T) method, which combines trace-norm regularization with LMNN, for person re-identification. The trace-norm regularization encourages the learned feature projection matrixes of low rank and thus controls the capacity of over-fitting. In addition, feature (attribute) bagging strategy is introduced to avoid dimension reduction, which may cause the loss of subtle feature information, and maintain the discriminative ability of image feature. Extensive experiments on two benchmarks demonstrate the superiority of proposed approach over state-of-the-art methods.