Contrast learning-based few-sample fine-grained image classification method

The invention discloses a few-sample fine-grained image classification method based on comparative learning, and belongs to the field of deep learning and computer vision, and the method comprises the steps: dividing an original data set, creating a meta-training set, a meta-verification set and a meta-test set, and for each task, dividing a support set and a query set from the meta-training set by using a sampling technology, a converter sharing weight is used as an embedded feature extractor of a support set and a query set; extracting embedded features of the support set and the query set from the converter, processing the embedded features to obtain feature representation for comparative learning, and after comparative learning features are generated, performing regularization processing on the features to remove the influence of data enhancement on the features; and then carrying out comparative learning, loss calculation, category prediction and model training. The method has high identification precisi