Clothes recommendation method and device based on interpretable convolutional neural network and terminal

The invention relates to the field of feature research about clothing recommendation, in particular to a clothing recommendation method and device based on an interpretable convolutional neural network and a terminal. The method comprises: carrying out the preprocessing of an image, and dividing the preprocessed image into a training set and a verification set; constructing a convolutional neuralnetwork, and constructing a feature matrix to extract clothing features of the acquired image; performing end-to-end iterative training on the convolutional neural network by using the training set; inputting the picture by a user into the trained convolutional neural network, outputting the feature map of the picture by the convolutional neural network, and recommending, by the convolutional neural network, clothes consistent with the feature map of the picture input by the user to the user. The features of the recommended clothes are visualized, so that a client can clearly know which kind of features on the clothes