Face Recognition and Identification Using Successive Subspace Learning for Human Resource Utilization Assessment

Attendance monitoring is an application of people counting with several use cases in various settings such as in corporate businesses and classrooms. Despite the vast number of possible applications, automated people counting or attendance monitoring applications in the Philippines are fairly uncommon. For example, establishments such as malls, supermarkets, terminals, and parks in the country still use manual counting. Their security personnel are given handheld counting devices to count the number of people entering the buildings or terminals. However, manual people counting, or attendance tracking methods are labor intensive, and thereby costly and time consuming. In spite of these issues, few establishments are interested in investing in automated counters as they tend to require devices or infrastructure (e.g., servers) which may be costly or impractical to scale. This paper aims to track a personnel's time management at work by monitoring his/ her dwell time in a dedicated workspace through face recognition. PixelHop algorithm was used to classify whether an individual is currently in their designated place, and to track the time they are within their workspace. This employs a successive subspace learning method which makes the face recognition more efficient with less overhead as compared to existing machine learning algorithms.