Target clothing image processing method based on generative adversarial network model

Abstract

The invention provides a target clothing image processing method based on a generative adversarial network model. The target clothing image processing method comprises the following steps: pairing a sample standard image with corresponding sample area images to form a sample pairing image set; optimizing loss function parameters of the generative adversarial network model according to the sample pairing image set; inputting the to-do area image into the generative adversarial network model, and outputting a template image; stretching and deforming the to-do area image to output a distorted image, so that the distorted image is aligned with the frame of the template image; and obtaining a pixel weight matrix, fusing the distorted image and the template image, and outputting a target clothing image. According to the method, the generative adversarial network model based on the perception loss function and the step-by-step image fusion technology are constructed, and the garment images with different angles and postures are converted into the target garment images with regular postures and enhanced textures to be searched and used by an intelligent system, so that the quality of the target garment images is improved, and the retrieval accuracy of the intelligent system is improved.

Description

technical field [0001] The invention relates to the field of image conversion, in particular to a method for processing target clothing images based on generative confrontation networks. Background technique [0002] With the continuous progress and development of Internet technology, the number of online shopping users is increasing day by day. Clothing accounts for a large proportion and revenue share among the many online products that can be browsed and purchased. In this context, how to efficiently search and recommend online clothing products has become an urgent problem to be solved. In order to accomplish the above tasks, a method that can effectively obtain the characteristics of specific clothing regions is needed to ensure the accuracy of subsequent correlation systems. [0003] Traditional methods generally use object detection technology to detect and cut clothing areas, and then perform further feature extraction based on the cut images, and then complete sub...

AI Technical Summary

Problems solved by technology

[0004] Based on this, the object of the present invention is to provide a target clothing image processing method based on the generation confrontation network model, through the generation confrontation network technology and the step-by-step image fusion technology to generate the target clothing image with texture patterns, to solve the problems caused by the target clothing image in the prior art. Intelligent system search and recommendation accuracy problems caused by lack of texture in clothing images

Images