Surakshavyuh makes surveillance smart and does in real-time what is usually a human’s job — monitoring CCTV camera footage for hours and alerting on suspicious movements
Screengrabs show how the technology can potentially capture human intrusion
Making CCTV surveillance smarter, a team from the Indian Institute of Technology-Bombay (IIT) has developed a technology, Surakshavyuh, that provides real-time analysis to alert security systems and can eliminate unrequired footage in offline scrutiny. The product has been deployed even on campus to aid in contact tracing and physical distancing amid Covid-19.
The technology has been developed by the National Center of Excellence in Technology (NCETIS) at IIT Bombay in collaboration with SrivisifAI Technologies Pvt. Ltd, a Pune-based start-up developing AI-enabled safety and security software solutions. It has been installed at Naval Dockyard Visakhapatnam and several other locations.
Professor Ganesh Ramakrishnan from the Department of Computer Science Engineering said that it includes features like real-time intrusion detection, perimeter monitoring, loitering detection, object tracking, face recognition, etc. Other solutions such as Jigyasa and Videosummy complement it with other features based on offline analytics. Jigyasa features a search platform on a video repository manager with features like text search, face search, etc. Another solution, Visiocity, condenses hours’ worth of video into a couple of minutes while preserving key events and vignettes from the original video and removing repetitive visual information.
Prof. Ramakrishnan added, “For any CCTV surveillance system, especially in the protected regions, the footage needs analysis — either real-time or for retrospective diagnosis. In the current process, in most cases, the onus of analysis is solely on the human viewing it. While we have researched and developed prototype solutions; the industry partner is looking at taking it to users.”
Professor Ramakrishnan said that the technology is unique in that it is grounded in Indian context which includes the need for resource-constrained, data-efficient machine learning algorithms and specific image and video scenarios with noisy backgrounds in the Indian context. The solution was developed on campus in five years and involved both field visits to ensure accuracy.