MUMBAI, India, April 10 -- Intellectual Property India has published a patent application (202441074675 A) filed by Impulse Compute, Bangalore, Karnataka, on Oct. 3, 2024, for 'image-to-text generation.'
Inventor(s) include Bantwal Harish Kamath; and Sparsh Jhariya.
The application for the patent was published on April 10, under issue no. 15/2026.
According to the abstract released by the Intellectual Property India: "Examples described relate to generating text from an image. In an example, a cloud computer system may receive an input document via a computer network. The system detects an object in an image present in the input document via a computer vision model. The system generates a first textual content of the image considering the object, via the computer vision model. The system then generates a second textual content of the image, based on the first textual content of the image, via a large language and vision model. The system generates a third textual content of the image, based on the second textual content of the image, via a large language model. The system applies a bias to the third textual content of the image. The system generates a fourth textual content of the image, based on the bias applied to the third textual content of the image."
Disclaimer: Curated by HT Syndication.
