Explore chapters and articles related to this topic
Current State of AI Governance & Regulation
Published in Catriona Campbell, AI by Design, 2022
However, Big Tech has begun to ask for regulation on facial recognition. This is seemingly a volte-face with Amazon, Microsoft and Google coming out in favour of legislating facial recognition technology. Critics say facial recognition systems are plagued by inaccuracies and have a detrimental effect on privacy. In the US, several cities and states have already banned their use. The US Government claims that the technology is plagued by inaccuracies, especially about identifying anyone who isn’t Caucasian. This potential racial bias appears to have stopped Facial Recognition in its tracks in the US. Faced with mounting criticism of its “Rekognition” system, Amazon company published “proposed guidelines” for the tech’s responsible use. Among the guidelines is a call for human oversight in the use of facial recognition systems by law enforcement and the argument that such tech should only be one of several different determinants in an investigation. The guidelines also clarify that Amazon supports transparency around the use of facial recognition systems by law enforcement. A representative said:
AI in Healthcare
Published in Shivani Agarwal, Sandhya Makkar, Duc-Tan Tran, Privacy Vulnerabilities and Data Security Challenges in the IoT, 2020
The field has become multifarious and scandalous as a great misuse of personal data occurred through Facebook. Despite all this, facial recognition will grow as the biggest field in the arena as the applications are hugely wide-ranging. The readability and accuracy of AI applications using facial recognition has benefitted a lot from the use of facial recognition in defense, at airports, and in various banking sectors where public access needs to be secure. Research into and the implementation of facial recognition will ultimately benefit defense and security agencies in tracking hackers, terrorists, criminals, and rogue elements in society; and businesses which need to provide more personalized data services to customers.
Blockchain-Based Regenerative E-Voting System
Published in Sudhir Kumar Sharma, Bharat Bhushan, Bhuvan Unhelkar, Security and Trust Issues in Internet of Things, 2020
S. Vishnuvardhan, R. Aswath Srimari, S. Sridevi, B. Vinoth Kumar, G. R. Karpagam
Election commission employees are given the privilege to oversee the voting procedure. Election commission employees will log in to the system with their electronic mail ID with a secure password. The voting system requires the facial recognition of the employees to allow access to the database. Facial recognition is one of the biometric software that maps individual facial characteristics mathematically and stores it in database as a faceprint. The individual is identified by matching the facial features captured using the camera with the faceprint stored in a database.
Gender and Age Classification Enabled Blockschain Security Mechanism for Assisting Mobile Application
Published in IETE Journal of Research, 2021
Sapna Juneja, Sourav Jain, Aparna Suneja, Gurminder Kaur, Yasser Alharbi, Ali Alferaidi, Abdullah Alharbi, Wattana Viriyasitavat, Gaurav Dhiman
As it is clear from the results that the proposed model has an accuracy of about 93%, which unlocks the application only when the algorithm confirms the person as an authorised user, which provides Confidentiality, Integrity, and Availability to our applications. Neither any user other than the actual user can access the data, nor are they able to make any changes to the confidential data, which provides an extra layer of security to applications and mobile devices. Along with this, the model also classifies the person’s gender as male or female and gives an estimation of their age. The concept of Facial recognition is used to first detect the human face in the image and then to recognise or confirm the person as authorised user. First of all, a dataset of 1000 images per person was captured using a webcam and a library called OpenCV. Haar Cascade classifier was used for face detection. Whenever the faces were detected according to the classifier, they were converted into grey scale and their resizing was done. LBPH model was used for face recognition and model was trained using the captured images after which the Face Lock section was ready to test in real-time. If the model’s prediction confidence is greater than 85% then only the applications are unlocked.
Autonomous Vehicles and Avoiding the Trolley (Dilemma): Vehicle Perception, Classification, and the Challenges of Framing Decision Ethics
Published in Cybernetics and Systems, 2020
Martin Cunneen, Martin Mullins, Finbarr Murphy, Darren Shannon, Irini Furxhi, Cian Ryan
Facial recognition (FR) technology has matured sufficiently to facilitate deployment in handheld applications ranging from mobile devices to security and surveillance systems. To do so, FR utilizes digital biometric or/and nonbiometric data and advanced machine learning pattern recognition algorithms such as Deep-Learning (Atallah et al. 2018) to provide an extensive review of face recognition comprising age estimation methods, databases, and algorithms, and including overall performance (accuracy rate) and limitations. While the drawbacks of such techniques rest on the fact that facial appearance may change with time (age), availability of data (images), and/or processing speed, data collation and computer processing trends are expected to overcome these limitations in the near future. Moreover, body shape recognition corresponding to fitness attributes (Awad et al. 2018) may also be classified.
Dangerous or Desirable: Utilizing Augmented Content for Field Policing
Published in International Journal of Human–Computer Interaction, 2020
Hendrik Engelbrecht, Stephan Lukosch
Although not a limitation of the current investigation itself, the ethical issues concerning the implementation of facial recognition should be mentioned. There is extensive and rapidly growing use of CCTV cameras in public places as well as body worn cameras used by police officers. Both technologies are already used in conjunction with human operators to monitor criminal activity and identify suspects. Despite substantial advances over the past years, the accuracy of automated facial recognition algorithms varies substantially between vendors (Grother et al., 2019). While this might not be a problem for forensic investigations, as human operators can utilize this to narrow down the potential pool suspects by matching faces to the police database, the usage in field-policing is more problematic. Field-policing represents a different interaction paradigm with facial recognition software. Decisions need to be immediate and involve direct contact with the potential suspect. There is little time for deliberation and the illusion of accuracy combined with time pressure might have devastating consequences. Many cities have actively prohibited the usage of facial recognition for the time being (Conger et al., 2019; Metz, 2019) . From the perspective of civilians, ethical considerations pitting privacy against security, especially for the usage of the technology in public spaces, is a major challenge for ethical implementation. This privacy-security balance, together with the issues of error rates and abuse or institutionalized expansion of use-cases, need to be carefully worked out before ethical deployment is possible (Brey, 2004).