智能虚拟试穿系统的设计与实现

Design and Implementation of Intelligent Virtual Try-on System

虚拟试穿在计算机领域和服装行业中都引起广泛关注,传统的虚拟试穿方法耗费大量计算人工成本和计算资源,研究人员将深度学习技术和虚拟试穿相结合出现了一些专业的虚拟试穿系统,但是这类系统比较关注于用户在目标服装下的试穿准确度,忽略了用户在不同姿势下的试穿效果。设计并实现智能虚拟试穿系统,旨在让用户可以选择不同的姿势,生成不同角度下的试穿效果。系统采用Django开发框架,通过调用之前提出的新的基于姿势引导下的虚拟试穿模型,实现不同姿势下的虚拟试穿,为用户在线上试穿提供更直观、更全面的结果;结合Vue框架实现前后端功能,存储并管理用户的相关数据。

Virtual try-on has attracted wide attention in both the computer field and the garment industry.Traditional virtual try-on method consumes a lot of calculation labor cost and computing resources.Researchers combine deep learning technology and virtual try-on technology to produce some professional virtual try-on systems,but such systems pay more attention to the accuracy of users'trying under the target garment and ignore the effect of try-on on the user in different positions.This paper designs and implements an intelligent virtual try-on system,which aims to enable users to choose different poses and generate fitting effects from different angles.The system uses Django development framework,and realizes virtual try-on under different poses by calling the new virtual try-on model based on pose guidance,which provides more intuitive and comprehensive results for users to try-on online,combining the Vue framework to realize the front and back end functions to store and manage the relevant data of users.