MUMBAI, India, March 13 -- Intellectual Property India has published a patent application (202641024802 A) filed by Sr University, Warangal, Telangana, on March 2, for 'a hybrid deep convolutional and instance segmentation model for intelligent video surveillance and automated suspicious activity detection.'

Inventor(s) include Mr. Gudupudi Ravi Kumar; Dr. V. Malathy; and Dr. K. Rajkumar.

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 discloses a hybrid deep convolutional and instance segmentation model for intelligent video surveillance and automated suspicious activity detection. The proposed system integrates high-speed object detection using YOLOv5 with precise pixel-level instance segmentation using Mask R-CNN to achieve accurate spatial localization of objects within complex surveillance scenes. A Pattern-Based Video Anomaly Detection Convolutional Neural Network (PBVAD-CNN) is incorporated to extract spatiotemporal features and analyze behavioral deviations across sequential video frames. The architecture performs trajectory tracking, motion pattern evaluation, dwell-time analysis, and contextual interaction modeling to identify anomalies such as intrusion, loitering, crowd congestion, unattended objects, and restricted area violations.The system further employs adaptive threshold optimization and context-aware calibration to minimize false alarms while maintaining high detection sensitivity. Edge-compatible deployment with model pruning and quantization ensures real-time inference with reduced computational overhead and low latency. Encrypted alert generation and secure event logging enhance operational reliability and forensic integrity. The modular framework supports scalable integration into transportation hubs, industrial facilities, smart cities, and critical infrastructure, thereby providing an intelligent, efficient, and robust automated surveillance solution for modern security environments."

Disclaimer: Curated by HT Syndication.