Abstract
Humans mostly use faces to identify/recognise individuals and the recent
improvement in the capability of computing now allow recognition and detection
automatically. However, there still exist quite a number of problems in the automatic
recognition of facial images. Histogram of Oriented Gradients (HOG) has been
recently adopted and seen as a standard for efficient face recognition and object
detection generally. In this paper, we investigate and discuss a simple but effective
approach to capturing student’s attendance register in a lecture hall by making use of
HOG features for detecting and recognising students face at different moods,
orientations, and illuminations. Our experiment detection and recognition output
show a good performance on our facial image database obtained from the University
of Johannesburg, this performance is due to HOG descriptors attributes which are
robust to changes in rotation and illuminations. Our system will help to save
instructional staff/lecturer time by eliminating manual calling of students name and
also help monitor students.