Unlock instant, AI-driven research and patent intelligence for your innovation.

A Appearance Attribute Modification Method of Apparel Images Based on Deep Learning

A deep learning and length attribute technology, applied in the field of image processing, can solve problems such as few clothing images, and achieve the effect of improving quality, accurate modification, and low impact

Active Publication Date: 2022-05-03
UNIV OF ELECTRONICS SCI & TECH OF CHINA
View PDF4 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, most of the existing work is done on face datasets, and rarely implemented on clothing images, which is a broader application scenario.

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • A Appearance Attribute Modification Method of Apparel Images Based on Deep Learning
  • A Appearance Attribute Modification Method of Apparel Images Based on Deep Learning
  • A Appearance Attribute Modification Method of Apparel Images Based on Deep Learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0018] The present invention mainly proposes a set of methods for migrating appearance attributes of clothing images based on generative confrontation networks. The details of the present invention will be described below in conjunction with the accompanying drawings.

[0019] The present invention proposes a method for modifying the appearance attributes of clothing images based on deep learning. The implementation steps of the method are as follows:

[0020] Step 1: For the clothing attribute dataset Shopping100k disclosed by Kenan Emir Ak in the paper, three sub-datasets are established according to the clothing appearance attributes. The clothing attribute dataset Shopping100k includes nearly 100,000 images, and each image is marked on 12 clothing image appearance attributes. Each clothing image appearance attribute has several categories. For example, the sleeve length attribute includes 9 possible Category: 3 / 4 length, Spaghetti (shoulderless), Sleeveless (sleeveless), ...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention discloses a method for modifying the appearance attributes of clothing images based on deep learning. The method mainly includes: obtaining sub-data sets of three clothing appearance attributes, namely color attributes, neckline style attributes and sleeve length attributes; establishing clothing image attributes based on deep learning Image appearance attribute modification model; use three sub-datasets to train the appearance attribute modification model of clothing image based on deep learning; use the trained clothing image appearance attribute modification model based on deep learning to modify the appearance attribute of the test clothing image. The present invention proposes a method for separating and expressing attribute coding and content coding. Using this method to migrate the appearance attributes of clothing images can improve the quality of generated clothing images and the success rate of appearance attribute migration, and at the same time ensure that the generated clothing images are consistent with Parts that are not related to the migrated appearance attributes remain as they are.

Description

technical field [0001] The invention belongs to the field of image processing, and in particular relates to a method for modifying appearance attributes of clothing images based on deep learning. Background technique [0002] Image editing and transformation has been a hot research direction in the field of computer vision. Being able to edit and migrate certain attributes in an image is very useful in certain scenarios, such as when a user is dissatisfied with a certain visual attribute in an image. For shopping websites, if users can modify the attributes of product clothing images with a low learning cost, it will undoubtedly greatly improve user experience. [0003] In recent years, deep learning and generative adversarial networks have developed rapidly. More and more image editing tasks are being done using Generative Adversarial Networks (GANs). However, most of the existing work is done on face datasets, but rarely on the broader application scenario of clothing i...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Patents(China)
IPC IPC(8): G06V10/44G06V10/56G06V10/764G06V10/82G06K9/62G06N3/04G06N3/08G06Q30/06
CPCG06N3/084G06Q30/0621G06V10/44G06V10/462G06N3/045G06F18/2132G06F18/241
Inventor 陈彦司新建胡洋
Owner UNIV OF ELECTRONICS SCI & TECH OF CHINA