计算机与数字工程2019，Vol.47Issue(4)：933-939,7.DOI:10.3969/j.issn.1672-9722.2019.04.038
基于改进的残差神经网络的服装标签属性识别
Recognition of Clothing Tag Attribute Based on Improved Residual Neural Network
摘要
Abstract
In recent years,the rapid development of clothing e-commerce has made the recognition of clothing tag attribute have a wide market demand,and the particularity of clothing images itself makes it more challenging to identify. In view of the great success of deep learning algorithms in many fields such as speech recognition and image processing,this paper designs a deep learn?ing model based on the residual neural network ResNet50,namely Res-FashionAINet,to apply it to the recognition of clothing tag attribute. Through three steps of data preprocessing,model training and attribute prediction,experiments are carried out on the FashionAI dataset composed of clothing pictures from real shopping platforms. The recognition accuracy rate is higher,and there is a certain advantage in the recognition of attribute tags for clothing images.关键词
服装图像/属性标签识别/深度学习算法/残差神经网络/FashionAI数据集Key words
clothing images/attribute tags recognition/deep learning algorithm/residual neural network/FashionAI dataset分类
信息技术与安全科学引用本文复制引用
张萌岩,何儒汉,詹伟,李敏,陈佳..基于改进的残差神经网络的服装标签属性识别[J].计算机与数字工程,2019,47(4):933-939,7.基金项目
中央级公益性科研院所基本科研业务费专项资金"微型消防站监测与应急指挥系统"(编号:C17392)资助. （编号:C17392）
