Object tracking algorithm of feature fusion Siamese network based on attention mechanism
An object tracking algorithm for feature fusion Siamese network based on attention mechanism is proposed.Aiming at the problem of insufficient feature robustness caused by the shallow depth of the feature extraction network of the object tracking algorithm,using the improved ResNet—50 network to extract the deep and shallow layers features of the template frame and search frame images,and using the channel and spatial attention mechanisms to fuse the extracted deep and shallow features.A template branch is added to the traditional Siamese network,and the first frame and the previous frame of the search frame are used as the object template to deal with the problem of template failure and tracking drift caused by the object tracking algorithm only using the first frame image as a template.Compared with the traditional classical tracking methods,the proposed algorithm has obtained the best tracking performance in the related experiments on the OTB100 and VOT2016 dataset,which verifies the effectiveness and feasibility of the proposed algorithm.
万方数据
维普
