Attendance Marking System using Periocular Recognition with Temperature Monitoring (ASPR)

Of late the COVID pandemic has necessitated authorities to make sure that every candidates are being worn mask. So, this forces to quit from conventional attendance marking systems which only relies on face recognition for biometric identification that involves close proximity or body contact. Wearing of masks can definitely occlude a major area of face. So the proposed project aims at using periocular recognition for biometric identification. The system proposed uses a pre-trained Convolution Neural Network (CNN) model that is VGG16 trained on ImageNet dataset to achieve the target of periocular recognition. Here it involves only a smaller region of interest and so external factors cause only less constraints to periocular recognition. Mask detection, which is an image classification phase, is done with MobileNet V2. This includes training which serialises face mask detectors to disk and deployment which outputs images as ‘with mask’ or ‘without mask’ [1] [2]. Non-contact IR sensor, MLX90614 IR sensor will automatically detect body temperature to determine whether the candidate's temperature is exceeding a threshold value.