MUMBAI, India, May 1 -- Intellectual Property India has published a patent application (202621023428 A) filed by Sunil Mane; Ayush Borhade; Amey Shinde; Onkar Shinde; and Pranav Kinge, Pune, Maharashtra, on Feb. 27, for 'ai-based multi-model intelligent theft detection and behavioural surveillance system.'
Inventor(s) include Sunil Mane; Ayush Borhade; Amey Shinde; Onkar Shinde; and Pranav Kinge.
The application for the patent was published on May 1, under issue no. 18/2026.
According to the abstract released by the Intellectual Property India: "The present invention, " AI-Based Multi-Model Intelligent Theft Detection and Surveillance System," provides a framework for automated theft detection and surveillance in retail environments. This system processes surveillance video through an integrated architecture consisting of video processing, detection layer, behavioural analysis and decision layer. The Video processing layer performs human pose estimation, retail object detection, safe zone detection and multi-person tracking. Detected objects are then dynamically associated with the tracked person based on predefined spatial proximity between object bounding regions and detected hand keypoints. The Behavioural Analysis Layer evaluates behavioural indicators which includes pickup, concealment, disappearance detection and abnormal posture probability. Then these indicators are combined using a weighted suspicion scoring mechanism which is expressed as linear combination of above given behavioural indicators. A memory-based stabilization module is then used to maintain the alert consistency, hence there by reducing false positives in the form of false alarms. Integrating this with safe zone logic helps to distinguish between legitimate item placement from concealment behaviour. As the calculated suspicion score exceeds the predefined threshold value, the system then generates automated alerts and visual indicators without any need of continuous human monitoring. The invention enhances detection accuracy, reduces false alarms and enables intelligent, scalable and real-time retail surveillance systems."
Disclaimer: Curated by HT Syndication.
