Han Dynasty Clothing Image Classification Model Based on KNN-Attention and CNN

Abstract

Aiming at the problem that the traditional clothing image classification model cannot effectively extract the instance information of other samples in the training set, a clothing image classification model of Han Dynasty based on KNN-Attention and CNN was proposed. Firstly, KNN-Attention was used to extract the clothing image information of K instance samples similar to the original training samples. Secondly, CNN is used to further extract the local key features of clothing images. Finally, the output information of KNN-Attention and CNN is integrated, so as to achieve the purpose of effectively utilizing the training set instance information in the task of clothing image classification. The experimental results show that the proposed model is better than the traditional classification model and can effectively improve the classification effect of clothing images.