AI-powered facial recognition puts 850 criminals on Kolhapur police radar during Navratri

Kolhapur superintendent of police Neelotpal said that a private agency is monitoring the temple premises live, with facial recognition enabled on cameras at key entry and exit points

With lakhs of devotees expected at Kolhapur’s Shree Ambabai Mahalaxmi temple during Navratri, the police are using artificial intelligence/AI-powered facial recognition technology to identify habitual offenders, including those from the neighbouring state of Karnataka, and prevent thefts in crowds.

The said system has helped prevent thefts, with the police reporting a significant reduction in cases in and around the temple. (REPRESENTATIVE PIC)

The police have uploaded the photographs of around 850 people with criminal records, including externed offenders and habitual property offenders, to a facial recognition system integrated with surveillance cameras at the temple. Alerts generated by the system will help police personnel in plainclothes to locate individuals for verification.

Kolhapur superintendent of police Neelotpal said that a private agency is monitoring the temple premises live, with facial recognition enabled on cameras at key entry and exit points. “Of the 125 cameras used for live monitoring, around 22 cameras at the entry and exit points are equipped with facial recognition. The system analyses images of people entering the premises and flags potential matches with those in the database,” he said. The local crime branch (LCB) provided photographs of around 850 people with police records, including externed offenders and habitual property criminals. LCB personnel in plain clothes have been deployed round-the-clock at the temple premises to respond to alerts, Neelotpal said.

The said system has helped prevent thefts, with the police reporting a significant reduction in cases in and around the temple, he said. The inclusion of offenders from Karnataka reflects the police’s focus on habitual thieves who may travel across the state border to target devotees. Pickpocketing and theft of mobile phones, wallets, jewellery and other valuables are particular concerns at the crowded religious site.

The West Maharashtra Devasthan Management Committee expects more than 1.5 million devotees to visit the temple over the nine-day festival. The crowds pose a challenge for conventional surveillance, and make it harder for the police to identify known offenders. The facial recognition system compares images captured by surveillance cameras with photographs in the police database. A potential match prompts personnel on the ground to locate the individual and verify their identity. An alert alone, however, does not establish that a person has committed an offence. The police are using the technology alongside conventional surveillance and deployment of personnel in plainclothes to identify suspects more quickly and prevent thefts during the festival.