IIT-Bombay develops AI platform for real-time video surveillance | Bombay News


State-of-the-art video surveillance platform developed by researchers at the Indian Institute of Technology-Bombay (IIT-B) has found application in military surveillance as well as remote monitoring of violations of social distancing standards in the middle of the Covid-19 pandemic.

The Surakshavyuh platform, originally designed in 2017, has now grown into an enterprise-grade video analytics solution based on machine-learning technology capable of detecting physical intrusions and wandering, monitoring perimeters and tracking objects. , counting crowds and recognizing faces, among other features. . It was designed as part of an industry collaboration between IIT-B’s National Center of Technology Excellence (NCETIS) and SrivisifAI Technologies Pvt Ltd, a Pune-based company that works to make technology intelligence artificial (AI) and affordable video analytics for mass consumption.

As the IIT-B team worked on the science behind the tools, SrivisifAI brought the product to market.

“For any video surveillance system, images must be analyzed – in real time or for retrospective diagnosis. We have developed these solutions with extensive field studies, research iterations, and developing the required algorithms that can alert end users of the products to take necessary action. It is even then that we are continuing our research to push the boundaries of the detection of human-object interactions, unusual events, etc. Said Ganesh Ramakrishnan, Institute Chair Professor, Department of Computer Science and Engineering, who led the project team.

Unusual human activity includes loitering, sneaking in, trespassing, entering and exiting the premises.

Surveillance currently depends mainly on images from CCTV cameras. “CCTV videos are generally subject to post-mortem event analysis. Our tool provides us with real-time analysis, ”Ramakrishnan said.

Besides Surakshavyuh, the team also offers Jigayasa, an offline analytics solution that works as a video repository and search platform with features like text search and face search. The two can also be used together.

As an example use case, templates can be used to issue a trigger warning if more than five people gather in one location or if one person is not wearing a mask, features that can be used in the pandemic. of Covid-19 to detect compliance with the rules in public spaces. A pilot project has been set up on the campus of the institute.

Subhasis Chaudhuri, Director of IIT-B, said: “Visual surveillance has become commonplace, whether in single-family homes or in public places. It is also very important to have good surveillance of vital installations. We are very pleased that IIT-Bombay has developed a very user-friendly and robust surveillance system, thanks to the efforts of our scientists. I hope that various business users will also benefit from this system.

The algorithmic and data benchmarking suite called Visiocity allows hours of video to be condensed into minutes in a domain-specific way, preserving key events and thumbnails of the original video and removing information. repetitive visuals. Ramakrishanan said: “There are three solutions to consider, but since all are interdependent or complement each other, industry partner SrivisifAI Technologies can offer products and services which are also combinations, like 3rdAI which is a combination of several solutions mentioned above. -above. This is exactly the uniqueness that NCETIS has facilitated – bringing universities and industry together as very strong partners. “

Traditionally, deep machine learning models are trained on large data sets and require a lot of computing resources such as multiple graphics processing units, both of which can be expensive. Surakshavyuh’s design is based on efficient machine learning of data – learning with frugal amounts of data and learning effectively, an attempt by IIT-B researchers to train cutting-edge models in resource-constrained environments while affecting minimum precision.

“Humans, based on conscious thought, can be fairer (if they decide to) and based on complex decision making, can be more precise than machines. Machines can be good at remembering, thanks to their memory and consistency in decision-making. The data efficient machine learning paradigm that underpins our democratized AI effort in our products is critical to making this collaboration between humans and the AI ​​engine, ”said Ramakrishnan.

Leave A Reply

Your email address will not be published.