MUMBAI, India, June 22 -- Intellectual Property India has published a patent application (202641050104 A) filed by Ms. M. A. Hemalatha; Ms. S. Kaviya; Ms. M. Kowsika; Ms. S. Lavanya; and Ms. S. Logeshwari on April 20, 2026, for Ai Forensic Image Generation And Recognition System.
Inventors include Ms. M. A. Hemalatha; Ms. S. Kaviya; Ms. M. Kowsika; Ms. S. Lavanya; and Ms. S. Logeshwari.
The application for the patent was published on June 12, 2026, under issue no. 24/2026.
Abstract: ABSTRACT OF THE INVENTION: The project titled "AI-Based Forensic Face Recognition System Using Sketch Matching" presents an intelligent system designed to assist forensic investigations by identifYing individuals based on facial images or sketches. In many real-world scenarios, law enforcement agencies rely on sketches of suspects rather than actual photographs, making identification a challenging and time-consuming task. This system addresses the problem by integrating artificial intelligence and image processing techniques to enable accurate and automated matching between sketches and stored facial data.The proposed system utilizes image processing methods to conveti input images into sketch representations using techniques such as grayscale conversion, inversion, and Gaussian blurring. This allows the system to handle both real images and sketches uniformly. The core of the system is a deep learning-based face recognition model that extracts unique facial features and convetts them into numerical embeddings. These embeddings are high-dimensional vectors that represent the distinctive characteristics of a human face and are used for comparison and identification .The system employs a structured database to store user information and serialized facial encodings, ensuring efficient data management and retrieval. During the recognition process, the input facial embedding is compared with stored embeddings using distance-based similarity metrics. The system then calculates similarity scores and ranks potential matches based on confidence levels, providing the most relevant results to the user. Additional processes such as serialization and deserialization are used to store and retrieve complex numerical data effectively within the database. A web-based interface developed using Flask enables user interaction, including authentication, image upload, and result visualization. The system is designed to operate in real time, providing quick and reliable identification results. It is scalable and capable of handling large datasets, making it suitable for practical deployment in forensic and security applications. The integration of attificial intelligence ensures robustness against variations in facial appearance, lighting conditions, and image quality.Overall, this project demonstrates the effective use of AI and computer vision in solving real-world forensic challenges. By automating the process of suspect identification using sketches, the system reduces manual effort, improves accuracy, and enhances investigation efficiency. It represents a significant step toward intelligent forensic systems and highlights the potential of AT-driven solutions in improving public safety and law enforcement capabilities.
Disclaimer: Curated by HT Syndication.
