MUMBAI, India, July 7 -- Intellectual Property India has published a patent application (202641059606 A) filed by Sugashini K; Dhanush G; Hariharan B S; Boopathy Velan K; and Abishek K on May 11, 2026, for Truesight:a Multi Model Framework For Ai Image Authenticity Detection.
Inventors include Sugashini K; Dhanush G; Hariharan B S; Boopathy Velan K; and Abishek K.
The application for the patent was published on July 03, 2026, under issue no. 27/2026.
Abstract: ABSTRACT The present invention relates to an Artificial Intelligence-based image analysis system designed for detecting Al- generated images and identifying visually similar objects within digital images. The proposed system, named TrueSight, combines semantic, spatial, frequency-domain, and residual noise-based feature extraction techniques to improve image authenticity analysis and object matching accuracy. The system utilizes CLIP-based semantic embeddings, EfficientNet-based visual features, Fast Fourier Transform (FFT)-based frequency analysis, and Spatial Rich Model (SRM)-based residual noise extraction. The extracted features are processed and classified using an XGBoost-based machine learning model. The invention further includes a Similar Object Detection module using SIFT feature matching, template correlation, andNon-Maximum Suppression (NMS) for object localization. A cache- based memory mechanism stores repeated detection mappings for efficient processing. The invention provides a scalable and web- based solution suitable for digital forensics, media verification, surveillance, and intelligent image analysis applications. The modular architecture of the system enables efficient integration of multiple computer vision services within a unified platform. The proposed invention also improves detection reliability by combining complementary feature extraction methods instead of relying on a single- domain analysis approach. Furthermore, the system supports real-time image analysis, efficient data handling, and extensible deployment for future Al-based forensic and visual intelligence applications.
Disclaimer: Curated by HT Syndication.
