MUMBAI, India, June 30 -- Intellectual Property India has published a patent application (202641075099 A) filed by K. S. R. M College Of Engineering on June 17, 2026, for Ai-Based Vision Surveillance Framework For Accurate Fire And Smoke Detection.
Inventors include Kanta Yamini; Shaik Imam Khasim; Katigandla Sowmya; Papi Reddy Devendra Rama Linga Reddy; and B. Manorama Devi.
The application for the patent was published on June 26, 2026, under issue no. 26/2026.
Abstract: ABSTRACT [0024] The invention presents an advanced real-time intelligent vision-based fire and smoke detection system that overcomes the limitations of conventional sensor-based alarms by leveraging deep learning and computer vision for rapid, accurate, and reliable hazard identification from video streams. This system addresses critical challenges such as delayed response, limited coverage, and frequent false alarms in traditional setups, offering a proactive solution for early fire detection in diverse environments. By integrating the powerful YOLOv8 object detection model with OpenCV for video processing, the system achieves high precision in distinguishing fire and smoke from misleading visual elements like sunlight, fog, or artificial lights. Trained on specialized datasets augmented with challenging hard-negative examples, the invention delivers robust performance with minimal latency, making it highly suitable for continuous surveillance applications. The results demonstrate exceptional detection accuracy, reduced false positives, and seamless alert generation, contributing significantly to enhanced safety, property protection, and emergency response efficiency in modern infrastructure.
Disclaimer: Curated by HT Syndication.
