Jarvis AI Transforms Crowd Management at Chinnaswamy After IPL Stampede Tragedy

The stampede in the victory of the RCB winning in IPL 2025 ended on a very tragic note, with 11 casualties caused by a stampede. This had raised serious concerns regarding the infrastructural feasibility of the Chinnaswamy Stadium. The renovation of the same witnessed the partnership between Staqu Technologies and the Bangalore flagship IPL team, RCB, in order to inculcate technological developments in transforming the old stadium into a state-of-the-art, modernised facility.

A video analytics system named Jarvi has been developed to monitor the real-time footage from 400-500 CCTV cameras to manage the crowd efficiently. The smart system has been equipped with the modernisation of manual tasks like the detection of potential hazards and restricted area intrusions, as well as violent behaviour detection from the massive crowd.

“The system continuously tracks crowd build-up and triggers alerts before critical thresholds are breached. Specific risk signals it monitors for include sudden crowd surges, restricted area intrusions, and violent behaviour patterns, shifting crowd management fundamentally from reactive to predictive," Atul Rai, Founder and Chief Executive Officer of Staqu Technologies, said.

Now comes the real question of impotence. Will the AI be able to differentiate between a celebration and a stampede? If yes, how?

The answer is given by the CEO himself. He has explained how this new Jarvis shall integrate the power of density estimation with activity recognition. The system has been designed to analyse specific anomalies of the crowd, such as the number of people present in a zone in real time, along with their behaviour or their state of motion, as well as rest.

Rai explained, “The power of this approach lies in its precision. A marathon can visually resemble a stampede, but the density and formation context tell a completely different story.” “It is only when an unexpected spike in density coincides with running activity in a given zone that the system flags a potential emergency, making it far more accurate than any single-model assessment,” he added.