Clothing deformation via multi-scale patch adversarial loss

Abstract

An embodiment of the invention relates to clothing deformation via multi-scale patch adversarial loss. Generating a synthesized image of a person wearing clothing is described. A two-dimensional reference image depicting a person wearing an article of clothing and a two-dimensional image of target clothing in which the person is to be depicted as wearing are received. To generate the synthesized image, a warped image of the target clothing is generated via a geometric matching module, which implements a machine learning model trained to recognize similarities between warped and non-warped clothing images using multi-scale patch adversarial loss. The multi-scale patch adversarial loss is determined by sampling patches of different sizes from corresponding locations of warped and non-warped clothing images. The synthesized image is generated on a per-person basis, such that the target clothing fits the particular body shape, pose, and unique characteristics of the person.

Description

technical field [0001] The present disclosure relates generally to the field of computing, and more specifically to clothing warping using multi-scale patch adversarial losses. Background technique [0002] With advancements in computing device technology, online shopping has become a more preferred mechanism for consumers than the conventional method of visiting brick-and-mortar retail stores. However, despite the convenience provided by online shopping platforms, consumers cannot perceive how a particular fashion item represented by a product image will actually appear when worn by the consumer. In order to provide consumers with a preview of how fashion items will appear, conventional virtual fitting rooms use three-dimensional models to estimate the consumer's body shape. Such conventional three-dimensional models can be generated using a variety of different methods, such as using a depth-sensing camera, using two-dimensional images to estimate the consumer's three-dim...

AI Technical Summary

Problems solved by technology

However, such image overlay methods fail to accurately represent important aspects of the consumer image such as the consumer's body posture, as well as important aspects of the product image such as deformation to the product material, visual pattern of the product, etc.

Therefore, conventional methods that only consider two-dimensional information tend ...

Images