MUMBAI, India, March 13 -- Intellectual Property India has published a patent application (202621010559 A) filed by Symbiosis International, Pune, Maharashtra, on Jan. 31, for 'multi-domain facial attribute translation system using generative adversarial network with style-content disentanglement.'
Inventor(s) include Dr. Shreyas Rajendra Hole.
The application for the patent was published on March 13, under issue no. 11/2026.
According to the abstract released by the Intellectual Property India: "The present invention relates to a multi-domain facial attribute translation system (100) using a generative adversarial network framework with style-content disentanglement for high-fidelity image synthesis. The system comprises a generator module (110) for synthesizing translated images, a discriminator module (120) for adversarial evaluation, a mapping network (130) for generating domain-specific style codes from latent vectors, a style encoder (140) for extracting style features from reference images, and an adaptive instance normalization module (150) for dynamic style modulation. The system enables simultaneous manipulation of multiple facial attributes including gender, hair color, age, and expression while preserving identity consistency through reconstruction loss mechanisms. The invention achieves superior perceptual quality with Frechet Inception Distance scores of approximately 28.3 and identity retention accuracy exceeding 92 percent, while operating efficiently on standard GPU hardware configurations."
Disclaimer: Curated by HT Syndication.
