HYDERABAD: Police use AI for Bonalu crowd control
HYDERABAD: City Police have deployed Artificial Intelligence (AI)-based systems for the first time to manage crowds during the Ujjaini Mahankali Bonalu festivities, Commissioner Sajjanar said.
About 2,500 police personnel have been deployed for security arrangements. With crowds increasing every year, police have installed AI-based crowd management and surveillance systems at the temple.
CCTV cameras equipped with facial recognition technology analyse faces in the temple surroundings. This helps identify repeat offenders, chain snatchers and pickpockets, officials said.
As part of AI crowd management, the system accurately counts devotees entering and exiting the temple using threshold mapping. If the crowd exceeds safe limits, police receive alerts and intervene to regulate movement and prevent stampede situations.
How the system works
AI cameras continuously monitor the temple premises and queue lines. Algorithms identify human heads, shoulders or body shapes to count people in each frame automatically.
Density estimation and heat mapping not only count individuals but also assess crowd pressure in specific areas. The system highlights crowd concentration using colour codes.
Areas with moderate crowds appear in blue or green. As density increases, colours shift to yellow and then red, indicating a potentially dangerous situation.
The system also tracks crowd movement patterns and walking speed. It detects if people are moving rapidly in one direction or if congestion is forming at any point.
సికింద్రాబాద్ శ్రీ ఉజ్జయిని మహంకాళి అమ్మవారి బోనాల బందోబస్తు చర్యల్లో భాగంగా మహంకాళి పోలీస్స్టేషన్లో తొలిసారిగా ఏర్పాటు చేసిన ఏఐ ఆధారిత ‘క్రౌడ్ మేనేజ్మెంట్ సిస్టమ్’ను హైదరాబాద్ సీపీ శ్రీ వీసీ సజ్జనర్, ఐపీఎస్ పరిశీలించారు. ఈ సందర్భంగా వ్యవస్థ పనితీరును అధికారులను అడిగి… pic.twitter.com/xVoyNQRRdG
— Hyderabad City Police (@hydcitypolice) August 2, 2026
What is AI-based crowd management
AI-based crowd management uses computer vision and advanced algorithms to monitor and assess crowd density in real time.
During large gatherings such as festivals and public events, the system identifies potential risks like overcrowding or stampedes in advance and alerts authorities.
For instance, if a queue line area can hold a maximum of 500 people, this limit is defined as a threshold. When the number crosses around 450, the system automatically sends alerts to the central control room.
Police can then temporarily stop further entry and take preventive measures to avoid overcrowding.
