USPTO Patent for Text-to-Image Generation Using Language Models
Summary
The USPTO has granted Salesforce, Inc. a patent (US12585919B2) for systems and methods related to text-to-image generation using language models. The patent describes a mechanism to integrate pre-trained language models into text-to-image generation models, enhancing image generation capabilities based on textual prompts.
What changed
The United States Patent and Trademark Office (USPTO) has granted patent US12585919B2 to Salesforce, Inc. The patent covers "Systems and methods for text-to-image generation using language models," detailing a mechanism to replace existing text encoders with more powerful pre-trained language models. This involves training a translation network to map features from the language model output into the target text encoder's space, preserving the language model's structure while enabling its use in text-to-image generation.
This patent grant is primarily of interest to technology companies and researchers in the AI and computing fields. While it does not impose new regulatory obligations or compliance deadlines on regulated entities, it signifies a development in AI technology that may influence future product development and intellectual property strategies. Companies operating in AI-driven image generation should be aware of this patented technology, particularly if their products utilize similar integration methods.
Source document (simplified)
Systems and methods for text-to-image generation using language models
Grant US12585919B2 Kind: B2 Mar 24, 2026
Assignee
Salesforce, Inc.
Inventors
Ning Yu, Can Qin, Chen Xing, Shu Zhang, Stefano Ermon, Caiming Xiong, Ran Xu
Abstract
Embodiments described herein provide a mechanism for replacing existing text encoders in text-to-image generation models with more powerful pre-trained language models. Specifically, a translation network is trained to map features from the pre-trained language model output into the space of the target text encoder. The training preserves the rich structure of the pre-trained language model while allowing it to operate within the text-to-image generation model. The resulting modularized text-to-image model receives prompt and generates an image representing the features contained in the prompt.
CPC Classifications
G06N 3/0455 G06T 5/70 G06T 2207/20084
Filing Date
2023-01-31
Application No.
18162535
Claims
20
Named provisions
Related changes
Source
Classification
Who this affects
Taxonomy
Browse Categories
Get Telecom & Technology alerts
Weekly digest. AI-summarized, no noise.
Free. Unsubscribe anytime.
Get alerts for this source
We'll email you when ChangeBridge: Patent Grants - AI & Computing (G06N) publishes new changes.