基于DCGAN算法的服装款式交互设计  

Garment style interactive design based on DCGAN algorithm

作　　者：任雨佳 陈璐[1] 陈郁 REN Yujia;CHEN Lu;CHEN Yu(School of Textiles and Fashion,Shanghai University of Engineering Science,Shanghai 201620,China)

机构地区：[1]上海工程技术大学纺织服装学院,上海201620

出　　处：《毛纺科技》2022年第3期93-97,共5页Wool Textile Journal

基　　金：上海市教育委员会东方学者项目(TP2017074)。

摘　　要：基于用户参与度及体验感的服装交互式设计逐渐成为服装数字化的研究热点。为了提升客户对服装款式的满意度,提出了一种基于DCGAN算法的小黑裙服装款式图生成方法。首先训练DCGAN算法生成清晰稳定的服装款式图;其次利用用户主观筛选得到的服装款式图更新训练数据,建立个性化服装款式库;最后通过生成的款式风格以及用户喜爱的款式图数量,评估该方法是否可以达到提高用户满意度的目的。结果表明,优化后的DCGAN服装款式生成系统可以生成清晰稳定且更加符合用户需求的服装款式图,为以消费者决策为导向的交互式智能服装设计平台的开发提供思路。Garment interactive design based on user participation and sense of experience has become the research hotspot of clothing digitization.In order to enhance customer satisfaction with clothing styles,a generation method of little black dress style based on DCGAN algorithm was proposed in this paper.First,clear and stable garment style was generate by training the DCGAN algorithm,then,the training data was updated by using the garment style obtained by user subjective selection,and the personalized clothing style database was established,finally,whether this method can achieve the purpose of improving user satisfaction was evaluated through the generated style and the number of style favored by users.The results show that the optimized DCGAN garment style generation system can generate a clear and stable garment style map more in line with user needs,which provides ideas for the development of interactive intelligent garment design platform oriented by consumer decision.

关 键 词：DCGAN 小黑裙 款式图 图像生成 个性化定制 

分 类 号：TS941.26[轻工技术与工程—服装设计与工程]