Object Tracking Algorithm for Siamese Network Combined with Channel Attention Mechanism

Abstract

As an important branch in the field of computer vision, object tracking has been widely used in many fields such as intelligent video surveillance, human-computer interaction and autonomous driving. Although object tracking has imposing development in recent years, tracking in the complex environment is still a challenge. Due to problems such as occlusion, object deformation, and illumination change, tracking performance will be inaccurate and unstable. In this paper, an object tracking algorithm for Siamese network combined with channel attention mechanism is proposed. Firstly, the Siamese network is used to improve the ability to discriminate features; secondly, the channel attention mechanism is introduced to design a cross correlation module DCAM (Depth-wise Cross-correlation with Attention Mechanism, DCAM), which pays more attention to the features that are beneficial to the tracking results; finally, the stochastic weight averaging method is used to train the network to further improve the overall performance of the tracker. Experimental results on public data sets show that the proposed algorithm has higher accuracy and more stable tracking performance in complex tracking environment