Abstract: The difficulty of cross-modal Person re-recognition is that there are both huge differences between modalities and intra-modal changes. For this reason, this paper proposes a cross-modal pedestrian re-recognition method based on convolutional attention mechanism and multi-loss combination. First, the first three convolutional layers of the Resnet50 network are used in the two branches to extract the characteristic shallow features of the pedestrian image, and then the convolutional attention mechanism module is embedded to suppress the extraction of irrelevant information such as color, and the middle layer features and the backbone of the branch are merged The final features acquired by the network improve the discriminative power of the acquired features. Finally, the two-way cross-modal triple loss loss and identity loss are used to jointly constrain the dual-stream network to accelerate the convergence of the network model and effectively deal with the differences between modalities and intra-class changes. Experimental results The method proposed in this paper effectively improves the accuracy of the cross-modal pedestrian re-identification problem.