Facial Recognition Technology: A Hornet’s Nest
Written by: Sanighdha, UILS, Panjab University, Chandigarh
“Any society that would give up a little liberty to gain a little security will deserve neither and lose both.”
The Electronic Frontier Foundation defines Facial Recognition System as, “A method of identifying or verifying the identity of an individual using their face. It can be used to identify people in photos, video or in real-time. Law enforcement may also use mobile devices to identify people during police stops.” It is quite perspicuous from the above definition that, Facial Recognition Technology or System (FRT/S) is a method of obtaining the very rudimentary and pivotal recognition marks of a person and stockpiling the same in a perdurable database. Prima facie, it might appear as another programme or technology of the State but; this seemingly innocuous scientific advancement does more harm than good, sans any data protection laws in the country. A developing nation like India availing an advance automation system like the one mentioned above, would undeniably compromise the personal data of millions of citizens. In light of the aforementioned premise, it becomes imperative to scrutinize the issue more scrupulously and suggest relevant solutions to the same.
Automated Facial Recognition and how it Works:
It was on June 28, 2019 that the National Crime Records Bureau released a Request for Proposal for an Automated Facial Recognition System (AFRS) to be used by police officers across the country. But before delving into the issues further, the functioning of AFRS must be understood. A leading Report on the same explains that firstly, the computer learns each specific face by training a particular algorithm which is commonly a deep neural network on a vast number of photos that have faces in known positions. The algorithm will try to identify the face each time it is presented and subsequently start recognising it in the right manner and as conditioned. This can be easily compared to Russian psychologist, Pavlov’s Classical Conditioning technique-where two stimuli are repeatedly paired: a response which is at first elicited by the second stimulus is eventually elicited by the first stimulus alone. The identification of a face as desired is the first step in the AFRS and is known as the Face Detection Step. The aforementioned initial procedure is followed by the Recognition Step, which is carried out via second deep neural network. The step simply plans to make the algorithm so conditioned that it is able to distinguish between two distinct facial structures. This is done vide a vector network guarded by numbers formed for each different face. Each face has a different and unique vector map, highly distinguishable from another structure. The software formed in this way is an extremely advanced one and technically suitable for real-time deployments. In crowd setups, the computer scans frames of video usually captured at crowd pinch points, such as entrances to cricket stadiums. Rest is totally dependent on effective conditioning of the aforementioned software since after scanning- vector is made for each face and thus whatever face is meant to be recognized by the police forces, is finally identified. Above mentioned is the basic outline of the technology but the success rate of the software (if deployed) is predominantly contingent upon widespread digital penetration as well social awareness of the same.
Integrating AFRS in the Indian Policing System:
The Automated Facial Recognition System Technology is meant to serve twin purposes i.e., recognition and authentication. According to the United Nations Office on Drugs and Crime Report of the year 2017, there are only 144 policemen governing 1 lakh citizens in India. Nations such as Argentina (803), Bhutan (536), USA (206) et al; are far ahead from India in the existing policemen to citizen ratio. The aforementioned data is a clear disposition of the fact that, the police and law enforcement agencies in India are way understaffed than previously thought. This non-existence of adequate public servants and law enforcement agents is a clear path to a looming ruination. It not only makes good governance seem completely impracticable; but equally rules out any viability of effective implementation of the existing legislations in the country. An understaffed and unequipped police force is an antithesis to the successful management of law and order in the society. Unequivocally, its non-viability for prevention and detection of crime can be easily measured. Indubitably, an efficacious support system for the police forces becomes inevitable.
At this juncture, the proposal for introducing AFRS by the NCRB seems a worthwhile suggestion. The NCRB suggestion further calls out for gathering CCTV footage and photographs from newspapers as well as other such sources. The Neural network of the AFRS will act as a biometric technology using distinctive features on the face, in order to identify and distinguish an individual from the others. Spotting an individual in real-time scenario by using the said technique will not only strengthen the overburdened police force but, will also prove as an operatively potent mechanism to curb criminal activities. The project also proposes integration of the existing criminal database with the information so collected via the AFRS system. It is said to be compatible with the biometric of the persons such as the iris and the fingerprints. The integration of the proposed mechanism with the already organised database systems such as-Crime and Criminal Tracking System (CCTNS- managed by NCRB), Integrated Criminal Justice System (ICJS), State Specific Database System (SSDS) and the Khoya-Paya Portal; will not only reinforce the handling of criminals and criminal cases but, will also consolidate and substantiate the prevailing- scattered crime management methodology. CCTNS is a pan-India integrated database on the criminal activities and suspects, alongwith FIR registrations, investigation reports and charge sheets filed in all police stations regarding a particular crime. It also assures to offer grievance redressal mechanisms and online tracking of cases by citizens, apart from other innumerable services. ICJS on the other hand is any computer system allowing judicial officers and law enforcement agencies to electronically access and share, any information on jurisdictional lines. The Khoya-Paya Portal is an online application developed by the Ministry of Women and Child Development and the Department of Electronics and Information Technology (DeitY) that allows active citizenry participation in reporting missing or found young children.
The Artificial Technology used in the AFR System in the form of neural networks, is poised to play a pivotal role in crime prevention and crime detection alongwith simplified verification process as well as eased-out procedure for retrieval and analysis of the possessed data. However, it must be noted that active incorporation of AFRS into the existing database of the country is not a nascent idea. Such systems are already functional at major Indian airports, such as the Delhi IGI Airport. The establishment of the technology at prominent places was carried out under the DigiYatra Initiative of the then government. The Telangana Election Commission also proposed usage of facial recognition system in the electoral process for curbing the menace voter impersonation. The Machine Learning of AFRS is an evolution of the basic facial recognition systems of the early 1960s. Determinatively, as a part of the police force, it will act as a force multiplier and proper utilisation of the technology will definitely wrap up cases in a time-bound manner. However, apart from having large-scale advantages, the AFR system has been attacked on myriad grounds and this raises serious concerns over the widespread deployment of the same.
Grounds of Concern:
It is a commonly accepted fact that the citizens of our nation want an act police force that is capable of solving matters in due time. Delay in criminal investigation is often considered as an impediment in the justice delivery mechanism and further leads upto violent mass movement over legal inaction. Conjoining technology and manpower is a brilliant antidote to the above mentioned crisis but, the concerns that are forever juxtaposed with the incorporation of a technology that demands mass surveillance must also be looked into. These include fundamental questions such as the source of collection of images that would ultimately create the desired database; the relationship that private entities (such as Twitter and Facebook) will share with the law enforcement agencies and the intimation of the same to the citizens using these mobile-based applications and; the issue of consent while sourcing out profiles of each and every person for central storage. While the aforementioned are the primal concerns with respect to the use of AFRS, the integrity of the database and its relevant security has also not been discussed in the proposal. This clearly leaves out the legally recognised ‘fundamental right to privacy’, out on a lurch. Noting the inconsideration so shown by the respective authorities, a well-charted out plan and strategy is the need of the hour to tackle any future issues.
On the other hand, there is no provision with respect to the usage of evidence so collected in the honourable Courts of our country. Until and unless the judiciary accepts to consider it as evidence and the legislature steps up to make a legal framework for the same, using AFRS is like shooting in the dark. The four –point principle or the Proportionality Test for chalking out the legal status, constitutionality and applicability of a State action laid down in the Aadhar Judgment must be considered in the present scenario as well. The mass surveillance carried out by AFRS is much more advanced than the CCTV footage and collection, as it seeks to make up a whole layout of a person’s facial features and thus poses a serious concern. The up-gradation of the whole system, if incorporated, requires a huge amount of money and the explicit dependence of India on foreign imports for such technologies is another economic issue. Another related dilemma is the high dependence on Cloud services for storing information and the respective insecurity attached to the same.
Recently, the European Commission has banned the indiscriminate usage of mass surveillance system via the AFRS and has called for strict regulations governing the trespass of individual privacy. The National Institute of Standards and Technology (NISDT- USA) in its Study in the year 2019 found that the misidentification rate of the facial recognition systems was almost 100% in some areas. The conundrums highlighted by the EU and the NISDT must be kept in mind while deploying the AFRS in India. Covering the loopholes while integrating the technology in the Indian criminal law enforcement system is an effective way of proceeding with the same.
It is a trite that any new technology or scientific advancement is filled with uncertainties and hidden dangers. However, a little bit sophistication and tough regulatory environment can weed out the same. In the absence of any data protection law, the probability of government abuse cannot be ruled out. Ensuring the passage of the dormant Data Protection Bill must be considered before issuing acceptance to the NCRB proposal. Proper IT training of police officers must be undertaken so as to make them sensitive to the idea of AFRS. Pilot projects must be carried out in the country before deploying the technology on a pan-India basis. Multiple testing of the system in areas of high criminal activity as well as in areas of low criminal activity to ensure reliability of the same can prove efficacious in a densely populated country like India. Conclusively, without any strict rules and regulations, the technology is nothing but a trap gate for privacy of millions.
Supporting one of the world’s biggest IT workforces, India must now look into effective allocation of the same. The research and development of the technology must be carried out by intricate scrutinisation on each step and each level must be tested several times. The privacy concerns of the citizens must be central to any step taken for law enforcement in the nation. Public comments must be invited on any future legislation that seeks to provide validation to AFRS. Last but not the least, awareness about the same must be dissipated so as to make the public aware and educated about the State actions undertaken in their interests.
 Street Level Surveillance, Electronic Frontier Foundation, https://www.eff.org/pages/face-recognition.  India’s Facial Recognition System- RFP, NCRB, http://www.ncrb.gov.in/TENDERS/AFRS/RFP_NAFRS.pdf.  Joe Gervais, How does facial recognition work, IoT, https://us.norton.com/internetsecurity-iot-how-facial-recognition-software-works.html#:~:text=A%20facial%20recognition%20system%20uses,it%20also%20raises%20privacy%20issues..  Saul McLeod, Pavlov’s Dogs, Simply Psychology (2018), https://www.simplypsychology.org/pavlov.html.  Jason Brownlee, A Gentle Introduction to Deep Learning for Face Recognition, Machine Learning Mastery (Mar. 31, 2019), https://machinelearningmastery.com/introduction-to-deep-learning-for-face-recognition/.  Chris Udermans, Facial Recognition firm can ID masked faces in crowd, TechNode (Mar. 10, 2020), https://technode.com/2020/03/10/facial-recognition-firm-can-id-masked-faces-in-a-crowd/.  Annual Report 2017, UNODC, https://www.unodc.org/documents/AnnualReport/Annual-Report_2017.pdf.  Id.  S. Jeyanthi, N. Uma Maheshwari, R. Venakatesh, Neural Network Based Automated Fingerprint Recognition System, https://www.researchgate.net/publication/282749559_Neural_network_based_automatic_fingerprint_recognition_system_for_overlapped_latent_images.  Crime and Criminal Tracking Networks and Systems, NCRB, https://ncrb.gov.in/en/crime-and-criminal-tracking-network-systems-cctns.  Integrated Criminal Justice System to be first launched in Telangana, Andhra Pradesh, NDTV (Jul. 17, 2016, 9:38 PM IST) https://www.ndtv.com/south/integrated-criminal-justice-system-to-be-first-launched-in-telangana-andhra-pradesh-1432557#:~:text=Hyderabad%3A%20Telangana%20and%20Andhra%20Pradesh,Madan%20B%20Lokur%20said%20today.&text=%22This%20system%20is%20going%20to,forensic%20science%20laboratories%20and%20Jails..  State Data Centre, Ministry of Electronics and Information Technology, https://www.meity.gov.in/content/state-data-centre.  Khoya-Paya, http://khoyapaya.gov.in/mpp/home.  Poulomi Saha, Delhi Airport to install Facial Recognition Technology, Trials to begin from September 6, India Today (Sep. 5, 2019, 18:17 IST) https://www.indiatoday.in/india/story/delhi-airport-facial-recognition-technology-trials-september-6-1595970-2019-09-05.  DigiYatra- A new Digital Experience for Air Travellers, National Portal of India, https://www.india.gov.in/spotlight/digi-yatra-new-digital-experience-air-travellers#:~:text=The%20'DigiYatra'%20is%20an%20industry,of%20passengers%20at%20the%20airports..  Telangana tests Facial Recognition in Local Polls as Privacy fears Mount, The Hindu (Jan. 22, 2020 14:16 IST) https://www.thehindu.com/sci-tech/technology/telangana-tests-facial-recognition-in-local-polls-as-privacy-fears-mount/article30623453.ece.  History of Facial Recognition Systems, Face First, https://www.facefirst.com/blog/brief-history-of-face-recognition-software/.  Justice K.S. Puttaswamy (Retd.) v Union of India (2019) 1 S.C.C. 1 (India).  Justice K.S. Puttaswamy (Retd.) v Union of India (2019) 1 S.C.C 1 (India).  Archana Chaudhary, India is Planning a Huge China Style Facial Recognition Program, The Economic Times (Sep. 20, 2019, 10:33 AM IST) https://economictimes.indiatimes.com/news/politics-and-nation/india-is-planning-a-huge-china-style-facial-recognition-program/articleshow/71213132.cms?from=mdr.  Akhilesh Shukla, Why Cloud Computing has high penetration in India, ET CIO. Com (Jun. 9 , 2014, 17:58 IST), https://cio.economictimes.indiatimes.com/news/cloud-computing/why-cloud-computing-has-high-penetration-in-india/30991526.  Facial Recognition-EU considers ban upto five years, BBC News (Jan. 17, 2020) https://www.bbc.com/news/technology-51148501#:~:text=The%20European%20Commission%20has%20revealed,for%20up%20to%20five%20years.&text=Exceptions%20to%20the%20ban%20could,well%20as%20research%20and%20development..  NIST Study Evaluates Effects of Race, Age, Sex on Face Recognition Software, NIST (Dec. 19, 2019) https://www.nist.gov/news-events/news/2019/12/nist-study-evaluates-effects-race-age-sex-face-recognition-software.  Personal Data Protection Bill, 2018, Ministry of Electronics and Information Technology, https://www.meity.gov.in/writereaddata/files/Personal_Data_Protection_Bill,2018.pdf.  Madhumita Jaganmohan, IT Industry in India-Statistics and Facts, Statista (Jan. 21, 2020) https://www.statista.com/topics/2256/it-industry-in-india/.
Opinions expressed in the blogs are the sole responsibility of the author(s) and do not necessarily reflect the views of The L Word Blog.