Is India ready for the implementation of the Automated Facial Recognition System (AFRS)? The answer to this question is another question – “At what cost?”
India’s law and order enforcement is a curious mix. At one end it is still governed by the colonial police law from 1861, at the other, it plans to implement the most advanced technology like AFRS across all agencies. This is a dangerous combination.
Sign up for your weekly dose of what's up in emerging technology.
The Personal Data Protection Bill 2019 is still under a joint parliamentary committee for review. There is a lack of the right laws to control who gets access and for what purpose. We’ve been working with facial recognition technologies long enough to understand the threat for potential misuse is very real. Before implementing any format of the facial recognition system, our country needs to safeguard its citizens. It needs to ensure that citizens feel safe and their personal space is not encroached upon.
The average Indian is afraid of the policeman. Several cases are under-reported because people are as afraid of the police as they are of the criminal, sometimes even more so. The courts are notorious for delaying cases. Eventually, attaining justice through the right means is an arduous journey. Law enforcement in India is impeded by a huge population, rampant corruption and a lack of trust between the police and the general public. The stress because of excessive workload does not make things easier for the police force either. The constitution of India gives the states and territories the powers to maintain law and order through their police forces. Every state has its own police force, making it difficult to track criminals when they cross over to different states.
Once deployed, AFRS will be the world’s largest facial recognition system. It will be a beneficial tool for the police forces, enabling it to track criminals across pan- India, using surveillance video from cameras placed at major tolls, railway stations, airports et al. For instance, recently, Rs. 250 crore was allocated from the Nirbhaya fund for installation of facial recognition cameras at 983 railway stations across the country.
The NRCB, in the first few pages of the RPF, has stated that “the Automated Facial Recognition System (AFRS) would help in automatic identification and verification of persons from digital images, photos, digital sketches, video frames and video sources by comparison of selected facial features of the image from an already existing image database”. A Facial Recognition System is an excellent investigation enhancer for identification of criminals, missing children/persons, unidentified dead bodies and unknown traced children/persons.”
What Is AFRS?
Automated Facial Recognition System (AFRS), to be used by police pan-India will be issued by the National Crime Records Bureau (NCRB). It would be a mobile and web-based application hosted in Delhi to help in crime prevention and detection, and fast track document verification. The AFRS is supposed to be interlinked with other existing databases like Crime and Criminal Tracking Network & Systems (CCTNS): managed by NCRB, Integrated Criminal Justice System (ICJS), State-specific database systems and the Khoya-paya portal.
AFRS will maintain an extensive database of photos and videos of peoples’ faces. In the second step, a new image of an unidentified person taken from CCTV footage will be compared to the database to find the correct match and identify the person. The system will add photos from newspapers, sketches et. al. and tag them for marks of identification like scars, tattoos and age. Using AI, it will be able to search with accuracy and speed to raise an alert if a blacklisted match is found.
How Does AFRS Work?
AFRS aims to provide a national-level searchable platform of facial images. The threat lies in the power that a facial recognition system provides to the ones with access. Deep learning-based technologies can provide real-time face detection that can quickly and accurately match images against massive databases.
It can coordinate between several law-enforcement agencies of the country and create a database of criminals that can be assessed by anyone with clearance. The definition of ‘potential criminal’, again is something that’s not very clear when it comes to technology like facial recognition. In the Delhi riots, the Delhi Police used video footage to identify protestors’ faces using facial recognition technology.
AFRS, once functional, will be deployed and utilised by both the State police, the Central forces and agencies working under the aegis of Central government. The NCRB states that the system will function as a repository of photographs of criminals present within the country. This would allow the policing agencies to detect crime patterns and understand the modus operandi of criminals across the states.
Proceed With Caution
India does not have a data protection law, yet. Law enforcement agencies need to have a high degree of discretion in the absence of these safeguards. There should be predefined, detailed processes to define and most importantly, limit the use of AFRS. Experts have voiced concern regarding the low accuracy of facial recognition algorithms when it comes to minorities, women and children. Before deployment, law enforcement agencies should understand its shortcomings and act accordingly.
Image recognition can be an ineffective and erroneous task in the early stages of a non-trained algorithm. The algorithm needs to be trained with the right data before being deployed in a criminal justice system. Marginal and vulnerable groups are more often over-represented, which make them potential victims of false positives. A compulsory training program for everyone with access to AFRS needs to be developed.
AFRS is a powerful tool. Due diligence is needed before it is made available to law enforcement across the country. Along with the technology, the people using it also should be ready.