A subtle allocation of Rs 699 crore in Karnataka’s latest budget to install 7,500 surveillance cameras in Bengaluru is likely to enable the state capital’s police to bring in facial recognition technology and video monitoring, officials said, even as experts flagged privacy concerns stemming from the possible machine-learning based plan.

While the project has been in the process for three years, Bengaluru Police officials said the latest funding announcement by the state government was a vital development.

A senior police officer said on condition of anonymity that the 7,500 cameras will include fixed day-and-night surveillance cameras, body-worn cameras and drones equipped with high-resolution image capturing devices. The police will also acquire facial recognition cameras and registration plate recognition devices. All cameras will be connected to a 40-seat Command and Control Centre, with two centres that will be mobile.

The 7,500 cameras are only the front-end of the intact project. As per another senior officer, close to 30% of the funds will be used to create a back-end. The officer said that to enhance  the CCTV network, the police will introduce technologies such as facial recognition.

“We want to have surveillance capability where a vehicle spotted at one traffic signal can be identified and tagged so that other cameras can trace the vehicle and predict its possible path. Similarly, we want to be able to track persons who are of interest using the technology,” the second officer added, asking not to be named.

He said that the proposed network will build the capability to go through recorded footage and identify patterns. “The technology will help us go through hours of footage and find patterns such as cars with a certain colour or all the buses that passed through one area.”

The officer elucidated that the system would depend on the footage collected from the CCTV cameras installed across the city.

“Apart from the CCTV cameras operated by the police, the Karnataka Public Safety Act gives us access to CCTVs of any establishment, which gets at least 100 visitors a day.

In the case of a situation where CCTV footage needs to be analysed in real-time or in retrospect, it will be done at the command centre. The same footage or analysis can be sent to respective police stations as well... using an internal network,” he said.

Vidushi Marda, a lawyer who works on emerging technologies, said that facial recognition and other similarly placed biometric applications increasingly seem popular with law enforcement agencies.

“As these systems contemplate collecting, analysing sensitive personal data, and basing consequential decisions on these inferences, keep in mind that these technologies are not silver bullets but rather fallible and often clunky in how they operate on the ground, lending a whole new layer of concerns in accuracy. In the absence of a data protection framework, this is a worrying prospect.”

“The assumption of the legality of these systems needs to be meaningfully scrutinised as well; in my reading, there is currently no legal framework from which these systems draw their legitimacy,” she said.

“Most recently, the Lucknow Police also announced plans to use emotion recognition technologies to identify ‘distressed’ women on the streets.”

A senior Bengaluru Police officer, however, said that the technology will be used only for detection and surveillance associated to a crime.