Siamese Object Tracking Algorithm Combining Residual Connection and Channel Attention Mechanism

Abstract

<p indent=0mm>Aiming at the problem that Siamese tracking algorithm is easy to track drift or loss in complex situations such as target deformation and similar object interference, a target tracking algorithm combining residual connection and channel attention mechanism is proposed. First, the shallow structure features and the deep semantic features extracted from the template branch network are effectively fused through residual connections to improve the model’s representational ability. Second, the channel attention module is introduced to make the model adaptively weighted to different semantic target feature channels to improve the generalization ability of the model. Finally, a weight mask based on correlation response values is designed and proposed to increase the weight of similar semantic target loss values during offline training, so that the model is enhanced discrimination of similar semantic targets in end-to-end offline learning. The results from comparative experiments with mainstream tracking algorithms on standard tracking datasets OTB, TempleColor128, VOT2016 and VOT2018 show that the algorithm is highly competitive in tracking accuracy and success rate, with superior real-time performance and reliability.