Facial recognition is a database matching system that uses an algorithm to identify a person based on a digital map of their face. There are two main applications for this technology:
The first is identification verification. In this, a face map is obtained to match a person’s photo in a database to validate their identity.
Second, there is identity 1-n identification. In which a face map is made from a photo or video and then matched with the whole database to find out who is in the photo or video.
What is the story so far?
According to information obtained through the Right to Information (RTI) Act by the Internet Freedom Foundation, a New Delhi-based digital rights organisation, the Delhi Police considers matches with a similarity threshold of 80% or higher created by the FRT system to be successful.
Why is the Delhi police deploying FTR?
The FRT was initially acquired by the Delhi Police to find and identify the missing children. According to RTI responses from the Delhi Police, the acquisition was permitted by a 2018 Delhi High Court judgement, Sadhan Haldar vs. Delhi NCT.
Available data suggest that the FRT was regularly used by the Delhi Police for investigations and during the northeast Delhi riots of 2020, the Red Fort violence of 2021, and the Jahangirpuri riots of 2022.
What makes FRT harmful to use?
Both the Union and State governments in India have rapidly deployed FRT in recent years without any legislation to regulate it. The use of FRT brings up two problems: problems with wrong identification because the technology isn’t very accurate, and problems with mass surveillance because the technology is often used in the wrong way.
The technology’s accuracy drops drastically by race and gender, according to a study. Either the person will be incorrectly identified as someone else, or they will not be confirmed as themselves, resulting in a false positive or false negative. There is a risk of prejudice being directed at the wrong person in cases of false positive results.