- Patent Title: Techniques for domain to domain projection using a generative model
-
Application No.: US17384357Application Date: 2021-07-23
-
Publication No.: US11880766B2Publication Date: 2024-01-23
- Inventor: Cameron Smith , Ratheesh Kalarot , Wei-An Lin , Richard Zhang , Niloy Mitra , Elya Shechtman , Shabnam Ghadar , Zhixin Shu , Yannick Hold-Geoffrey , Nathan Carr , Jingwan Lu , Oliver Wang , Jun-Yan Zhu
- Applicant: Adobe Inc.
- Applicant Address: US CA San Jose
- Assignee: Adobe Inc.
- Current Assignee: Adobe Inc.
- Current Assignee Address: US CA San Jose
- Agency: Kilpatrick Townsend & Stockton LLP
- Main IPC: G06N3/08
- IPC: G06N3/08 ; G06F3/04845 ; G06F3/04847 ; G06T11/60 ; G06T3/40 ; G06N20/20 ; G06T5/00 ; G06T5/20 ; G06T3/00 ; G06T11/00 ; G06F18/40 ; G06F18/211 ; G06F18/214 ; G06F18/21 ; G06N3/045

Abstract:
An improved system architecture uses a pipeline including a Generative Adversarial Network (GAN) including a generator neural network and a discriminator neural network to generate an image. An input image in a first domain and information about a target domain are obtained. The domains correspond to image styles. An initial latent space representation of the input image is produced by encoding the input image. An initial output image is generated by processing the initial latent space representation with the generator neural network. Using the discriminator neural network, a score is computed indicating whether the initial output image is in the target domain. A loss is computed based on the computed score. The loss is minimized to compute an updated latent space representation. The updated latent space representation is processed with the generator neural network to generate an output image in the target domain.
Public/Granted literature
- US20220122221A1 TECHNIQUES FOR DOMAIN TO DOMAIN PROJECTION USING A GENERATIVE MODEL Public/Granted day:2022-04-21
Information query