摘要
随着电子商务的发展,网络购物已成为选购服装商品的主要渠道,使得服装图像的准确分类变得愈发重要。近年来,对服装图像分类的研究大多专注于服装种类,而对服装风格识别的研究相对较少。随着线上服装交易量的不断增加,各类平台累积了大量无法得到充分利用的未标注风格的服装图像。针对这一问题,本文运用基于Inception v3的迁移学习技术,构建服装风格识别模型,并通过Deepfashion服装数据集对该模型进行重训练,生成服装风格分类器,而后将此服装风格分类器应用于对服装图像数据集进行批量标注。
With the development of e-commerce,online shopping has become the main channel for people to buy clothing products,and accurate classification of clothing images has become very important.In recent years,researches on clothing image classification have mostly focused on clothing types,while clothing style recognition research has been relatively rare.With the continuous progress of online clothing transactions,various platforms have accumulated a large number of unlabeled clothing images that cannot be fully utilized.In response to this problem,this paper uses transfer learning technology to build a clothing style recognition model based on Inception v3,and uses Deepfashion The clothing dataset is retrained to generate a clothing style classifier.Finally,this clothing style classifier is applied to batch label the clothing image dataset.
作者
张艺凡
刘国华
盛守祥
王国栋
王力民
ZHANG Yifan;LIU Guohua;SHENG Shouxiang;WANG Guodong;WANG Limin(School of Computer Science and Technology,Donghua University,Shanghai 201620,China;Huafang Co.,Ltd.,Binzhou Shandong 256602,China)
出处
《智能计算机与应用》
2020年第5期14-17,23,共5页
Intelligent Computer and Applications
基金
科技部国家重点研发计划(2017YFB0309800)
上海市工业互联网创新发展专项项目(2019-GYHLW-004)。
