A Review of Secure and Privacy-Preserving Hand Gesture CAPTCHA Systems with Enhanced Liveness Detection

Text-and/or image-based CAPTCHAs, as a dominant way of differentiating bots from humans, are now often defeated by computer-Vision & opticalcharacter-recognition models prompting development of other verification techniques that are not easily automatable yet still convenient to humans.This review article presents a new option based on the user performance of randomly picked gesture in front of his/her regular webcam while he/she is under live human examination with just few extracted hand-landmarks.Prior work related to this area has looked at implications for spoofing resistance, usability, and privacy aspects.It has been pointed out that a gesture-based CAPTCHA could also be considered as a spoofing detection in videoand-image-based tasks.This is why the paper also examines different spoofing detection methods that are effective for both task types and evaluates whether they can be used for end users without the requirement to handle or store video data.The paper draws together past work on gesture-based CAPTCHAs, handlandmark recognition, face/gesture spoofing detection, together with an update on recent public availability and vulnerability disclosure of hand-gestor reCAPTCHA.It emphasizes the accuracy of recognition, capability of spoofing detection, factors on usability and privacy issues of the methods considered with a view to pointing out what remains to be done both on research level and real-life application to create a more pleasant and still beneficial user experience of privacy-and computationpreserving local processing.