A Target Tracking Algorithm Based on Channel Attention and Spatial Attention

Abstract

Due to issues such as target occlusion and complex background interference, the accuracy and stability of target tracking still need to be improved. To address these issues, this paper proposes a target tracking algorithm based on channel attention and spatial attention. This algorithm applies channel attention and spatial attention mechanisms to the field of target tracking. Among them, the channel attention mechanism enhances the ability to extract target features by focusing on important channels in the input feature map; The spatial attention mechanism achieves accurate estimation of target position by focusing on important areas in the feature map, enhances the expression ability of target features by utilizing channel attention mechanism, and suppresses interference from background features by utilizing spatial attention mechanism. Through experimental verification on the got-10k dataset, the algorithm proposed in this paper has higher tracking accuracy and stability in complex scenarios.