ABSTRACT
Authentication is the process of identifying and verifying the identity of an individual via different means. Authentication is an important step in every field that requires security, authenticity, permissions, and restrictions. In the field of technology where the end user is behind a screen and the means of identifying and verifying are limited, it becomes a challenging task to complete this process correctly with the resources available. Authentication can be as easy as providing a name, PIN, or password to more complex and secure method like OTP and Biometrics. Biometric Authentication has become mainstream and readily available since the launch of the first fingerprint scanning smartphone, but it still requires components like a fingerprint scanner that is not available with every digital device whereas cameras are more widely, easily available making it possible to implement Facial Recognition. Facial Recognition is the process of detecting and identifying face(s) from a still image, video, or live stream, it is a fast, reliable, and easy to use biometric that can be implemented on any digital device capable of capturing images. Using Facial recognition in combination with traditional username and password would create a Multi Factor Authentication that can be implemented in various scenarios. In this paper, We used our previous research and analysis of facial recognition techniques [1] to create a proof-of-concept attendance marking system that uses the camera to detect and recognise faces and marking attendance in real time using DLib [7] and facial recognition [2] libraries and created a machine learning module that use HOGs and facial landmarking techniques to create a facial profile of the input image. This system is based on computer vision and image processing and utilises the computing power of the processor as well as the GPU if available, to process the image for detection of faces and later recognising the faces from the database.
