BENGALURU: In a first-of-its-kind initiative, Bengaluru Police successfully deployed Artificial Intelligence (AI) to identify and track firecracker violations across the city during Deepavali. Officials said the system detected more than 2,000 incidents of firecracker bursting over five days, with 41% of them occurring after the 10 pm deadline mandated by the Supreme Court.
AI-powered surveillance deployed across high-density zones
The initiative made use of Bengaluru’s Safe City surveillance network, which includes thousands of cameras and an Integrated Command and Control Centre. A Firecracker Detection Video AI app analysed live feeds from over 200 cameras placed in congestion-prone areas such as Srirampuram, KR Market, HSR Layout, Haralur and Marathahalli.
Joint Commissioner of Police (Administration) Kuldeep Kumar Jain said the AI model was trained to detect visual indicators associated with crackers.
“The AI algorithms were designed to identify firecracker flashes, smoke and crowd hyperactivity in real time. Once detected, alert packets with location, timestamp and visual proof were instantly sent to the Command Centre via WhatsApp,” he explained.
Instant alerts help officers act faster
Earlier, police action depended largely on citizen complaints. By the time Hoysala teams responded, violators often dispersed. This year, the AI module enabled immediate alerts to field units, allowing faster deployment and preventing escalation.
“With the AI module integrated into our camera network, Hoysala teams receive alerts the moment an incident is detected. This helps us act quickly and prevents further bursting of crackers,” Kuldeep added.
The zones selected were primarily those with high footfall and past complaints of late-night crackers. Police said the technology not only helped enforce the 10 pm rule but also allowed them to proactively curb noise and air pollution, which typically spike during the festival.
Reduction in pollution and manual dependency
Authorities noted that the system significantly reduced dependency on public reporting and human monitoring. The use of AI-enabled classification also improved accuracy, ensuring that officers were not deployed unnecessarily.
Police said early analysis suggests that the number of large-scale cracker violations after 10 pm dropped compared to previous years — a direct result of quicker response times and heightened visibility of enforcement.
Bengaluru may expand AI-led enforcement
Following the success of this trial, Bengaluru Police is considering expanding AI for other violations such as illegal racing, chain-snatching detection, and crowd density monitoring during major events. Officials believe the Safe City AI integration can eventually become a citywide standard for real-time law enforcement.
The initiative is part of a broader push by Karnataka to incorporate AI into governance, including recent developments such as the launch of the AI-ready KEO PC with built-in agent BUDDH.
As Deepavali celebrations continue to evolve, police say AI-assisted surveillance will play a critical role in maintaining safety, reducing pollution, and ensuring compliance with legal restrictions.