Clothing attribute extraction using convolutional neural networks

Abstract

Automated annotation of cloth images is an appealing technique, which have many applications, such as cloth search and classification. This study suggest an automated cloth attribute annotation system suggested that extracts low-level features from images and learns multiple classification models for predicting 25 cloth attributes. This study uses a deep convolutional neural networks (CNN) algorithm for building classifiers. Our research results show that CNN-based approach outperforms the previous approach.