Method for classifying similar faces of different countries based on deep learning fusion model

The invention particularly relates to a method for classifying similar faces of different countries based on a deep learning fusion model. The method comprises the following steps: inputting face images into a trained face classification model to obtain predicted country types; during training, a face image is input into a trunk feature extraction network constructed based on fusion of a ResNet network and an SE module, and a fine-grained feature map is extracted; inputting the fine-grained feature map into a full-connection layer to obtain a predicted nationality type; calculating corresponding global loss; inputting the fine-grained feature map into a mutual channel loss function module to obtain local region loss; and mixing the global loss and the local region loss to obtain corresponding model loss, and optimizing parameters of the face classification model. According to the method, the ResNet network and the SE module are fused to learn higher-level feature representation and capture fine-grained featur