Research On Image Style Transfer Method For Clothing Creative Design

As the earliest category involved in e-commerce,clothing has become the largest and mature industry.Online clothing sales have many advantages that traditional models do not have,allowing users to fully enjoy the fun and interactive experience of online shopping.As the core technology of the virtual fitting system,clothing migration technology has attracted more and more attention.However,the difficulty of virtual fitting system is how to help users quickly and accurately find personalized clothing and artificial intelligence-assisted clothing design.We propose a study of style transfer method for clothing creative design to assist designers in designing fashionable clothing and provide users with personalized digital clothing customization.our method can solve the problem of not considering the texture,pattern,color,and material of clothing during style transfer.Such as clothing creative design elements,resulting in the loss of clothing structure and other issues,a better migration result is obtained,which provides a new idea for artificial intelligence-assisted clothing creative design.First of all,in view of the loss of clothing structure due to the lack of consideration of clothing texture,pattern,color,material,etc.,we propose a multi-style fusion fast clothing migration deep network model.In our model,we use more than one element.Firstly,perform semantic segmentation on the input clothing content image to extract the contour structure of the clothing content image.Then build a multi-style fusion pre-training network to fuse multiple clothing elements to form a new clothing style.Secondly,the extracted clothing contours and fusion styles input to the constructed clothing migration network.We optimize the network according to the characteristics of clothing content,style,color,and contour.Finally,we generate multiple customizations of clothing by applying different styles,and output the clothing migration results after contour enhancement.Then,in view of the above problem that the current clothing generation does not combine clothing style and style creativity.Therefore,a fast clothing generation network that combines creative design and style transfer is proposed.This method first inputs two clothing style images for component segmentation and semantic segmentation.We obtain the clothing component map and use the image fusion algorithm to fuse the images to obtain the clothing style image.Then,we input the clothing style map into the clothing migration network to generate the clothing migration image.Finally,we reconstruct the clothing migration image with high resolution to obtain high-quality clothing front image.Finally,in view of actual application scenarios and user needs,the above algorithms are combined to design and implement a prototype system of clothing creative design migration.the prototype system realizes the integration of multiple styles of clothing,creative design of clothing,migration of clothing styles,and mask generation of clothing images.The system consists of a multi-style generation module;a style migration module;a clothing style merging module and mask generation.Film generation module,and save the result,it is composed of the button to browse the picture.The interface consists of 4 module buttons to select images and generate images.