Abstract: Aiming at the problem that traditional content-based clothing classification had high requirements for image features, and its accuracy was difficult to meet the application requirements of clothing classification when there were many clothing styles, we proposed a  parallel self-attention classification network based on deep learning methods. The network added a parallel self-attention compensation branch  on the basis of ResNet50, which could improve the quality of feature extraction in clothing classification tasks, and gradually supplement shallow detail information missing from  deep network. A comparative experiment was carried out on the DeepFashion dataset, and the experimental results proved the effectiveness of this method.

Key words: clothing category classification, deep learning, self-attention mechanism, information compensation