Person re-identification methods based on the attention mechanism make more use of the first-order information in the feature,ignoring the second-o rd er information in the feature,and the correlation and fine-grained information between feature maps cannot be mined.This paper proposes a pedestrian re-iden t ification algorithm based on the second-order mixed attention module (SMAN) .Th e second-order mixed attention module (SOMA) is composed of the second-order ch an nel attention (SOCA) and the second-order spatial attention module (SOSA).This method imbed the Global Coordination Pooling function into SOCA and SOSA modules to learn second-order information in features.The SOCA module uses second-ord er information to learn the correlation between feature maps,While the SOSA modul e assignes weights to feature maps again,focusing on the fine-grained informat i on of feature maps spatial domain.The Rank-1accuracy of SMAN algorithm on mark e t-1501and DukeMTMC-ReID datasets reached 94.30% and 87.1%,respectively,and mA P reached 85.7% and 74.53%,respectively.Finally,the class activation map is u sed to verify the effect of SOMA module.Experiments show that SMAN algorithm ma kes full use of the second-order information in the channel domain and spatial d omain of the feature map.The performance of the algorithm is better than some ex isting person re-identification methods based on attention mechanism,and even close to the performance of some state-of-the-art methods.