Nation-Wide Facial Recognition

Indian child labour activist, Bhuwan Ribhu, points out there are over 3,00,000 missing children in the country. And there are over 1,00,000 children in various institutions. Matching them manually is obviously a practical task. In July this year, he launched a pilot program to match the digital databases using facial recognition technology. The outcome was very good:

“We were able to match 10,561 missing children… They are currently in the process of being reunited with their families.”

Given the severe shortfall in police (144 officer per 1,00,000 citizens) as opposed to, say, the EU (318 per 1,00,000), India needs to try and find different ideas to bridge the gap. Including, as it turns out, turning to facial recognition technology.  Delhi adopted it in 2018, along with Andhra Pradesh and Punjab.

And now the central government wants to expand the scope:

“It wants to construct one of the world's largest facial recognition systems. The project envisions a future in which police from across the country's 29 states and seven union territories would have access to a single, centralized database.”

What would be the source of photographs for this database? Mugshots of criminals, obviously. Also artist sketches of suspected criminals. Add to that passport photos and images from the increasing number of CCTV’s.

As you might have imagined, the idea is to expand the usage beyond just the police:

“Security forces would be equipped with hand-held mobile devices enabling them to capture a face in the field and search it instantly against the national database, through a dedicated app.”

In October, bidding was opened for this project. Pre-bid meetings with prospective vendors in July included discussions on topics like “whether it should be able to identify people who have had plastic surgery”. Given the lack of experience in such matters amongst Indian companies, we might need to tap into foreign companies. A local player has been mandatory in whatever consortium is picked. But that raises concerns of national security and espionage by those foreign companies.

Experts believe this is going to be a very challenging project, especially the centralized database aspect. On the other hand, we did build the Aadhar system, so who knows? Others say the number of CCTV’s are too small today to make such a project meaningful. Then again, once you have such a system, CCTV feeds could be made mandatory in more places, and it may get added in phases into more and more public places. London and Shanghai have lots of such feeds, but not the whole of UK or China, so perhaps we’ll start with the major cities too.

Delhi, for examples, started the project to install 3,30,000 new cameras across the city. They cite improving the safety of women as the driver. The government plans to add such feeds in all stations in the country by 2021.

There are concerns over lack of data protection laws in the country. Privacy concerns are another issue. And as always, there is the fear it would be used against minorities. And if it were to get linked to Aadhar, it might end in a “total, permanent, surveillance state”, worries Apar Gupta from an NGO. So yes, it is going to need a balancing act:

“Caught between the need to improve its policing outcomes and to protect its citizen's privacy, India will be walking a tightrope when it comes to building its national facial recognition database.”

I think we need to be pragmatic: we are not rich like the West, we don’t yet have all the right laws on privacy, but staying put and doing nothing until those things happen first is not an option either. So we move along, and adjust along the way.