Emerging TechnologiesFuture of WorkInnovationPrivacy
Surveillance in the Dark: Use of Thermal Facial Recognition
These days, there is a trend to install CCTV cameras as if it is a panacea to all problems. CCTV cameras are being installed with full vigor not only at every nook and cranny of every street of the country but also in closed spaces like classrooms and train coaches.
But it is difficult, rather nearly impossible, for law enforcement agencies to monitor all recordings at all times manually. To deal with this impracticability, CCTV cameras are increasingly being complemented with facial recognition technology which would help identify suspects and suspicious activities automatically. Nevertheless, usual CCTV cameras have their own limitations and anti-social elements can easily deceive them in night-time or when lighting is not proper, or by wearing mask or face covering makeup. As images in these cases would not be clear, it would be difficult for the facial recognition software to identify the persons in them. Like in JNU violence incident, the offenders were wearing masks, making their identification an arduous task for the authorities.
To deal with these shortcomings, the National Crimes Record Bureau (NCRB) is planning to install thermal facial recognition cameras as well. NCRB has released tender, and request for proposal for Automatic Facial Recognition System (AFRS). Proposed centralized AFRS will lay the groundwork for “a national level searchable platform of facial images” and will create an archive of photographs that can be used to identify and track criminals. The database will be created using passport, Crime and Criminal Tracking Network and Systems (CCTNS), Interoperable Criminal Justice System (ICJS) and Prisons, Ministry of women and child development (KhoyaPaya) State or National Automated Fingerprint Identification System or any other image database available with police/other entities.
Thermal cameras and facial recognition systems can be very effective in overcoming some of the major problems associated with the use of normal CCTV cameras. The thermal or infrared (IR) camera, captures the heat emitted by humans and forms an image (called thermogram); and therefore, the lighting of the area, makeup, mask, and, in some cases, even plastic surgery cannot hamper the capturing of a thermal image of a person. Nowadays, thermal cameras are being used in China to find persons with high temperature who could be potential coronavirus suspects. With thermal cameras, they can be tracked even if they are wearing masks.
This is not to suggest that the thermal facial recognition system is without limitations. Though it is easier to capture a thermal image, it is difficult to identify a person using thermal image because images in the database with which a person’s thermal image is to be matched contain mostly visible (non-thermal) images. To tackle this, various methods are being developed and used, like facial recognition using a fusion of thermal and visible images; converting thermal image to visible image and then matching the visible image obtained, and using unique deep neural networks or vein structure of face for the thermal facial recognition. Further, the thermal image is influenced by the emotional and medical state of a person.
Studies have shown that the notion of technological objectivity of facial recognition system and AI in general, is a myth, and facial recognition system reflects the same ‘biases’ as humans because the technology is trained on data that is labeled/classified by humans. In the Indian context, the installation of AFRS would work against tribals, scheduled castes, and people from minority groups as NCRB’s 2018 report itself says that about two-thirds of Indian prisoners (both undertrial and convicts) are from these communities. If trained on existing data that is highly skewed against tribals, scheduled castes, and people from minority groups, AFRS would certainly make their life more miserable than before as it would be more difficult to prove the biases of AI/AFRS.
The installation is problematic on the legal side as well. In response to legal notice by Internet Freedom Foundation (IFF), NCRB stated that its tender for AFRS is based on a 2009 Cabinet Note. Cabinet Note lacks statutory backing and does not satisfy the legal requirement of a three-part test recognized by Puttaswamy judgment.
Thermal data, unlike facial images, is not expressly recognized by the definition of “biometric data” under Personal Data Protection Bill, 2019; still it can be considered biometric data under “any other similar personal data resulting from measurements or technical processing operations carried out on physical, physiological…” of a person. As biometric data is sensitive personal data, facial images and thermal data shall be accorded higher protection. Mere notice that CCTV cameras and facial (or thermal facial) recognition technology are installed at a place would not constitute ‘explicit consent’ for the collection of facial and thermal data. However, the Central Government has the power to exempt certain entities from this requirement in the interest of public order and national security under the 2019 PDP Bill. This exception, if generously used, would effectively impinge on the freedom of speech and association, leading to self-censorship and creation of police state to the likes of China and Russia.
Before the installation of facial recognition systems, it is important to study the feasibility and practical impact of such technology on civil liberties. This concern is all the more apt for thermal facial recognition technology which still needs to be meticulously scrutinized through the lenses of technical efficiency and accuracy, and human rights impact. Further, the thermal data of a person shall be expressly included in the definition of biometric data in the final version of the PDP Bill. At least until a robust data protection law is in place, NCRB and other authorities like Railways which are deploying facial recognition, shall not do so in haste without properly evaluating its drawbacks and consequences on civil liberties. As once installed for whatever noble cause, they would always remain susceptible to be misused for stifling dissenting voices.
 –, Delhi school becomes first ever to provide live CCTV video feed to parents: CM Arvind Kejriwal, India Today, Jul. 08, 2019, https://www.indiatoday.in/education-today/news/story/delhi-school-becomes-first-to-provide-live-cctv-video-feed-to-parents-cm-arvind-kejriwal-1564401-2019-07-08.
 –, Railway Board chairman says CCTV in all stations and coaches by March 2022, Live Mint, Dec. 31, 2019, https://www.livemint.com/news/india/railway-board-chairman-says-cctv-in-all-stations-and-coaches-by-march-2022-11577762567176.html.
 Advertisement for Automated Facial Recognition System, National Crimes Records Bureau, http://ncrb.gov.in/TENDERS/AFRS/Advt-Automated Facial Recognition System (AFRS)v2.pdf.
 Request for Proposal for Automated Facial Recognition System, National Crimes Records Bureau, http://ncrb.gov.in/sites/default/files/tender/RFP_NAFRS.pdf.
 M. Krišto & M. Ivašić-Kos, An Overview of Thermal Face Recognition Methods, https://bib.irb.hr/datoteka/941518.5074-An-Overview-of-Thermal-Face-Recognition-Methods-v2.pdf.
 Reuters, China firm develops a system to recognize faces behind coronavirus masks, CNBC, Mar. 09, 2020, https://www.cnbc.com/2020/03/09/china-firm-develops-system-to-recognize-faces-behind-coronavirus-masks.html.
 Steve Lohr, Facial Recognition Is Accurate, if You’re a White Guy, New York Times, Feb. 08, 2018, https://www.nytimes.com/2018/02/09/technology/facial-recognition-race-artificial-intelligence.html.
 Shemin Joy, Majority prisoners in Indian jails are Dalits, Muslims, Deccan Herald, Jan 01, 2020, https://www.deccanherald.com/national/north-and-central/majority-prisoners-in-indian-jails-are-dalits-muslims-790478.html.
 NCRB finally responds to legal notice on facial recognition, we promptly send a rejoinder…, Internet Freedom Foundation, Nov. 08, 2019, https://internetfreedom.in/the-ncrb-responds/.
 2017 SCC Online SC 996, Justice K.S.Puttaswamy(Retd) v. Union Of India And Ors., 494 W.R. (Civ.) 2012, Supreme Court, decided on 24 Aug. 2017.
 Personal Data Protection Bill, 2019, Bill No. 373 of 2019, S. 3(7), http://126.96.36.199/BillsTexts/LSBillTexts/Asintroduced/373_2019_LS_Eng.pdf.
 Id., S. 11.
 Id., S. 35.
 IP based Video Surveillance System being installed by Indian Railways at railway stations to enhance security, Press Information Bureau, Jan. 08, 2020, https://pib.gov.in/newsite/PrintRelease.aspx?relid=197330.