As India faces a steep rise in cases, the need for quarantine facilities has also risen sharply and subsequently, the need for technology solutions to help maintain strict quarantine protocol has grown. A team consisting of Prof Ganesh Ramakrishnan and more than 15 other individuals who are part of a startup, SrivisifAI Technologies Pvt Ltd, that works in the field of Artificial Intelligence / Machine Learning / Deep Learning (AI/ML/DL) enabled Video Analytics. The startup has developed Suraksha Vyuh, a Real-Time Analytics product that uses Analytics on CCTV footage to enhance the level of security provided by CCTV cameras – both in real-time and post facto analysis. The work is largely part of the National Centre for Excellence in Technology for Internal Security (NCETIS)

How does the product work?

The system uses real-time analytics to keep track of any unexpected movements in the CCTV feed such as camera tampering, masked face detection, unusual object detection, inactivity detection etc.  The solution was motivated by two persistent issues, inattention caused by exhaustion and difficulty in having ex post facto analysis. It solves these issues by providing features like summarization, real-time alerts, smart search etc.  

An overview of the architecture of the Surakhsavyuh world has been provided in the flowchart below.

Uses – 

Suraksha Vyuha (SV) solution was developed to carry out real-time analysis of video feed from the CCTV camera stream using a massively scalable architecture that supports both on-premises and cloud deployment. A suitable adaptation of SV (called Drishti) is being commercially deployed for compliance and teaching quality monitoring for a flagship programme of the Ministry of Rural Development (MoRD). SV is actively used at IIT Bombay, and its effectiveness was realized during the ongoing COVID-19 situation. It is being used in IITB for quarantine surveillance and lockdown of 100+ workers. The team has provided an evaluation version of SV to a few private/public organizations. The team has also developed specific solutions for automatic number plate recognition (ANPR) and scene text recognition in general. There is also a human-machine interaction framework built around ANPR and the work which is being adopted commercially through their industrial partner AIFY Labs (http://www.aify.co.in/)

Scope of Expansion – 

The team is planning to deploy at facilities that have similar requirements for safety and surveillance. The team is developing customized solutions of AI/ML/DL-enabled VA capabilities to serve specific needs of public/private organizations. The team has developed another product Jigyasa which is a solution meant for processing CCTV footage offline. Jigyasa is now being taken up for massively distributed computing over GPUs. 

With AI being used in multiple fields, such innovations can potentially revolutionize criminal investigation and security services. With the capacity of the health authorities and police being stretched to the brink, Real-Time Video Analytics could solve the problems regarding the security of rapidly increasing quarantine facilities. With the team working on customized solutions, it won’t be long before such systems are available in many everyday facilities as well.

The content on this website is strictly the property of Insight and the Students’ Gymkhana IIT Bombay. If you wish to reproduce any content herein, please contact us:
Chief Editors: Amogh Gawaskar and Suman Mondal

Mail to: insight@iitb.ac.in