Since last week, the Madurai City Police in Tamil Nadu have been using a facial recognition application in a bid to nab potential criminals roaming around, especially in the night. The police claimed that with the rising number of chain snatching, mobile snatching, burglaries, and robberies, it has “given importance to vehicle checking, patrolling in crime prone areas and crime prone hours of the day”. However, when we reached out to Madurai City Police Commissioner Prem Anand Sinha, asking why only a facial recognition app can perhaps help them, he declined to comment. News agency ANI first reported the Madurai City Police using FaceTagr.
This isn’t the first time that a police department has used FaceTagr’s facial recognition tool. The Chennai police has been using it for several years, and had also caused controversy when it was seen scanning faces of people protesting against the controversial Citizenship Amendment Act earlier this year. Apart from that, NotionTag Technologies, the parent company of FaceTagr, had also developed a facial recognition based quarantine monitoring app which was being used by several police stations in Tamil Nadu.
How the police plans to use the app
Sinha informed MediaNama that earlier, the police used to carry out fingerprint matching of people it suspected. However, that was a long and cumbersome process. “How do you know if someone roaming around in the night is a criminal? To reduce the harassment faced by people [who’re not criminals], we decided to start using this app”, Sinha told us.
The police have access to a facial database of around 3,000 pictures, taken since 2016, and another dataset of city criminals, which is to become the basis to carry out face matches. However, the police will keep expanding this particular database, and not just from images captured in Madurai, but also neighbouring districts. “The database will be updated on a regular basis as and when crime cases are reported. We are also talking to neighbouring district police in this regard so that their criminal database can also be created and integrated with the existing one,” Sinha said.
Photos of the “criminals” have been uploaded to a central database and the police personnel who want to check the criminal database can either take a live picture of an individual or can use an existing photo available in their phone’s gallery and run it through the database.
The app also shows a “percentage of matching score”. “This way the works of checking criminal record of a person becomes very easy and effective. Using only 50 kb of data for each photo it can work in even slow internet connections and give results in less than a second,” the police said.
When we asked whether there were any procedural safeguards in place, to ensure that the police on the ground doesn’t misuse the app, we were told that the app can only be accessed by few select police personnel who have been given a one-time access code, which is non-transferable. The police on the ground has also been instructed not to use the app “indiscriminately”, but mainly to check people in a vehicle in the night and when there are grounds of suspicion. Sinha also told us that the Commissioner’s office can track any photos that might have been taken using the app, and if they find that it has been misused, they will take action.
How the app works
To understand how the tech behind the app works, we reached out to Vijay Gnanadesikan, co-founder and CEO of FaceTag, who explained to us that the facial recognition algorithm a “confidence score” of the match upon verification, which “augments the manual verification process”. When we asked what a “confidence score” means, he said that the app will basically show that there is “90% confidence that the two pics are of the same person”.
He claimed that the accuracy of the facial recognition system is 99.4%, but when we asked how the company managed to develop an algorithm which such a high rate of accuracy, he declined to comment, saying that the information was confidential. “But it is not from any data from any of our customers. I will assure you that,” he said. He also told us that all the data captured by the police is stored on a “government certified server”, without specifying whether these are government-owned servers, or merely certified by the government.
When asked about the average time it takes for the app to return with a search result, Gnanadesikan said that it mostly depends on the Internet speed. However, at the same time, he claimed that face comparison takes “only less than 40ms”, and typically less than a second to show the result.
Facial recognition everywhere
The Indian government, several law enforcement agencies, and a number of state governments have deployed, or are planning to deploy facial recognition surveillance tools. The National Crime Records Bureau is currently working towards building a national level facial recognition system, and only very recently revealed that it wants to tests the system on mask-wearing faces, and for it to generate “comprehensive biometric reports”.
In Telangana, police have been known to demand people to undergo a facial recognition verification exercise, randomly, claiming that they do so to nab criminals. Like how police in Tamil Nadu have FaceTagr, Telangana’s police use an app called TSCOP. The state is also planning to introduce facial recognition for obtain ration, and is already using it for issuing renewed driving licenses and pensions. Telangana is also the first state in the country to pilot facial verification of voters during its civic polls earlier this year.
The Vadodara City Police, in Gujarat had earlier told MediaNama that it was planning to use Clearview AI’s controversial facial recognition software in public places to track “property offenders”. This software could also be used in CCTVs installed at “specific locations” in the city. The Indian Railways said that it was in the process of installing Video Surveillance Systems (VSS), equipped with a facial recognition system, in 983 railway stations across the country. A number of airports in the country have introduced facial recognition based boarding solutions.
It is worth noting that Indian agencies are installing facial recognition surveillance tools when India doesn’t have a data protection law. Moreover, the Personal Data Protection Bill, 2019, which is currently being deliberated upon by a Joint Parliamentary Committee, has carved out exemptions for government agencies to adhere to provisions of the Bill. This suggests that government institutions like the NCRB, or the Railways could potentially be able to collect, store and process biometric data of Indians without necessarily adhering to the provisions in the Bill.
- How Telangana is using the pandemic to push facial recognition tech on its population
- Interview: Telangana could soon use facial recognition authentication for ration distribution, says state’s IT secy Jayesh Ranjan
- India’s NCRB to test automated facial recognition system on ‘mask-wearing’ faces
- Exclusive: Concerns around number of active users, and ‘backdoors’ raised at an NCRB facial recognition meeting