MUMBAI, India, Feb. 13 -- Intellectual Property India has published a patent application (202641008990 A) filed by Vemu Institute Of Technology, Chittoor, Andhra Pradesh, on Jan. 29, for 'system and method for ai-based forensic face reconstruction from textual description.'
Inventor(s) include Dr. P. Nirupama; Ms. A. Shailaja; and Ms. B, Bharathi.
The application for the patent was published on Feb. 13, under issue no. 07/2026.
According to the abstract released by the Intellectual Property India: "This invention introduces a new system and process for reconstructing forensic facial images from descriptive text in the form of artificial intelligence through the use of Deep Convolutional Generative Adversarial Networks (DCGAN). The system overcomes the drawbacks of conventional suspect identification techniques like eyewitness drawings and database-based facial recognition. Using a pretrained DCGAN model that was fine-tuned on the CelebA dataset, the invention generates realistic facial images that relate to descriptive features given in natural language. The process includes preprocessing text inputs to obtain semantic features, projecting the features onto latent space vectors, and creating corresponding facial images using the trained generator network. The architecture is designed with Python, Flask, and Anaconda, with modular sub-systems for data import, model inference, and image rendering. It accommodates integration with law enforcement databases and forensic software and can be deployed in scalable form in investigative processes. This invention improves the speed, accuracy, and objectivity of suspect visualization, particularly in photogram-negative cases. It is cost-effective, operationally viable, and can be modified to fit different forensic scenarios. The architecture of the system caters to future upgrades like attribute editing, multi-modal input, and real-time synthesis. In total, it provides an innovative forensic facial reconstruction technique that closes the gap between visual identification and descriptive evidence."
Disclaimer: Curated by HT Syndication.
