MUMBAI, India, Jan. 2 -- Intellectual Property India has published a patent application (202541122096 A) filed by Vellore Institute Of Technology, Vellore, Tamil Nadu, on Dec. 4, 2025, for 'adaptive text-guided image editing using diffusion-based learning and forgetting mechanisms.'
Inventor(s) include Dr. Prakash G; Pranav Handa; Harsh Kumar; and Anshuman Das.
The application for the patent was published on Jan. 2, under issue no. 01/2026.
According to the abstract released by the Intellectual Property India: "The present disclosure provides a computer-implemented method for adaptive text-guided image editing. The method includes receiving an input image and a text prompt describing desired modifications, encoding the input image into a latent representation using a variational autoencoder encoder, generating a source text embedding from the input image using an image captioning model, fine-tuning selected diffusion model parameters by jointly optimizing the source text embedding and selective neural network parameters while implementing forgetting mechanisms, generating a target text embedding from the text prompt, combining the source and target text embeddings using vector projection or subtraction to create a blended embedding preserving image identity while incorporating desired modifications, applying the diffusion model with the blended embedding to iteratively denoise the latent representation, and decoding the denoised representation using a variational autoencoder decoder to produce an edited output image."
Disclaimer: Curated by HT Syndication.
