Sampling latent variables to generate multiple segmentations of an image

US11430123B2 · kind B2 · utility

Assignee

• DeepMind Technologies Limited · GB

Inventors

• Simon Kohl · London, GB
• Bernardino Romera-Paredes · London, GB
• Danilo Jimenez Rezende · London, GB
• Seyed Mohammadali Eslami · London, GB
• Pushmeet Kohli · Bellevue, US
• Andrew Zisserman · London, GB
• Olaf Ronneberger · London, GB

Key dates

• Filing date: May 22, 2020
• Grant date: Aug 30, 2022
• Expiry date: Sep 8, 2040

Classification

• Technology area (CPC G): Physics
• CPC primary: G06T2207/30096
• WIPO field: Medical technology
• WIPO sector: Instruments

Abstract

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for generating a plurality of possible segmentations of an image. In one aspect, a method comprises: receiving a request to generate a plurality of possible segmentations of an image; sampling a plurality of latent variables from a latent space, wherein each latent variable is sampled from the latent space in accordance with a respective probability distribution over the latent space that is determined based on the image; generating a plurality of possible segmentations of the image, comprising, for each latent variable, processing the image and the latent variable using a segmentation neural network having a plurality of segmentation neural network parameters to generate the possible segmentation of the image; and providing the plurality of possible segmentations of the image in response to the request.