Artificial intelligence model training for image generation

By parsing static images into content items and structured representations, and training an image design model with high-quality data and noise techniques, the method enhances the quality and diversity of generated images, addressing inaccuracies in current generative AI models.

WO2026049733A1 Publication Date: 2026-03-05GOOGLE LLC
1 Cites 0 Cited by

Patent Information

Application Number
PCT/US2024/044413
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Filing Date
2024-08-29
Publication Date
2026-03-05

AI Technical Summary

Technical Problem

Current generative artificial intelligence models produce inaccurate and low-quality images, lacking robust training data and often incorporating hallucinations and artifacts, which affects the quality and diversity of generated images.

Method used

A method involving an image parser model to parse static images into content items and structured representations, combined with an image design model trained using high-quality static images and noise techniques to generate diverse and accurate training data, resulting in higher quality image designs.

Benefits of technology

The approach generates more diverse and higher quality images by expanding training data through noise techniques and using structured representations, reducing model complexity and computational resources while improving accuracy and relevance.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for training AI models to generate images. In one aspect, a method includes obtaining first training data including a set of first training samples. Each first training sample includes a first image. An image parser model is trained using the first training data to output structured representations of input images. Second images are processed using the image parser model to generate second training data including a set of second training samples. Each second training sample includes a second image and a corresponding structured representation of content items depicted by the second image output by the image parser model based on the second image. An image design model is trained using the second training data to output a structured representation for an input including a set of content items.
Need to check novelty before this filing date? Find Prior Art