Govt’s approach to recognition is faulty & pushed by assumptions that are faulty.
Police in Chennai is based to help maintain order and law in areas of the city. In the same way, Punjab Artificial Intelligence System, a recipient of a FICCI smart policing award, uses facial recognition for identification. Authorities in Hyderabad is using facial recognition to identify’persons of interest’ with footage.
The Indian government has also announced plans to an overarching national Automated Facial Recognition System (AFRS), which is utilized for’criminal identification, verification and its dissemination among various police organisations and units throughout the country’. Pared down to principles, facial recognition technology of this kind captures faces in public areas, generates a unique biometric map of every face (like a fingerprint or DNA), and then contrasts these maps to present databases like the Crime and Personal Tracking Network and Systems (CCTNS), as well as live CCTV footage if such is the requirement. A number of assumptions guide this turn towards recognition in India, every one of which I unpack below.
Is this effective?
It is assumed that the use of facial recognition will present speed and efficiency into one of the most understaffed police forces in the world. On the other hand, the logical jump from usage of technologies to efficacy is a costly one. In 2018, Delhi Police reported that the facial recognition system on trial was operating at a precision rate of 2 percent. In 2019, once the precision rate fell to less than 1 percent, the Ministry of Women and Child Development reported that the machine could accurately distinguish between boys and girls.
It is crucial to note that accuracy rates of facial recognition software globally have been lowest for racial minorities, children, women, and non-binary genders. These biases make it a application in law enforcement, since people anticipated to be identified by a recognition system and the overlap between groups from the criminal justice system is important. Institutional and historic discrimination will be consequently exacerbated by the uncritical adoption of those systems against illegal targeting of individuals and illegal arrests. As of this moment, efficiency is an unfulfilled promise, however, unreliability and discrimination are demonstrated effects.
Privacy and surveillance concerns
But even perfectly precise facial recognition systems are problematic. A second premise that guides police deploying facial recognition from India is that the use of those systems does not increase privacy or surveillance concerns. It’s argued that facial recognition will be used by law enforcement locate missing kids and to monitor criminals. In context of this AFRS, it is also argued that using facial recognition only adds’another information coating to investigation by allowing matching picture of suspects or lost persons together with CCTNS”s photo database.
This premise is untenable. From each individual who’s recorded by the camera to recognise suspects or’ persons of interest’, they need to accumulate, store and analyse data for recognition methods to operate in the context of crime prevention. In other words, if a system must carry out the job of recognition in such use cases, it must sort a game . Authorities deploying facial recognition must reckon that however laudable the goal of recognition systems may be surveillance is a necessity for them.
The claim that facial recognition adds a layer of advice for picture matching is an oversimplification. This technology involves creating a biometric map of every person’s face, based on the distance between their eyes, nose, mouth, and jaw, the size of their forehead etc.. It is far more similar to a fingerprint or DNA evidence or an iris scan than it’s fundamentally, and to a picture affects how law enforcement agencies identify people.