Invention Publication
- Patent Title: Image Enhancement via Iterative Refinement based on Machine Learning Models
-
Application No.: US18155420Application Date: 2023-01-17
-
Publication No.: US20230153959A1Publication Date: 2023-05-18
- Inventor: Chitwan Saharia , Jonathan Ho , William Chan , Tim Salimans , David Fleet , Mohammad Norouzi
- Applicant: Google LLC
- Applicant Address: US CA Mountain View
- Assignee: Google LLC
- Current Assignee: Google LLC
- Current Assignee Address: US CA Mountain View
- Main IPC: G06T5/00
- IPC: G06T5/00 ; G06N3/08 ; G06N3/045 ; G06T5/50 ; G06T3/40

Abstract:
A method includes receiving, by a computing device, training data comprising a plurality of pairs of images, wherein each pair comprises an image and at least one corresponding target version of the image. The method also includes training a neural network based on the training data to predict an enhanced version of an input image, wherein the training of the neural network comprises applying a forward Gaussian diffusion process that adds Gaussian noise to the at least one corresponding target version of each of the plurality of pairs of images to enable iterative denoising of the input image, wherein the iterative denoising is based on a reverse Markov chain associated with the forward Gaussian diffusion process. The method additionally includes outputting the trained neural network.
Public/Granted literature
- US11756166B2 Image enhancement via iterative refinement based on machine learning models Public/Granted day:2023-09-12
Information query