부산대학교 도서관

The computer vision has been advancing rapidly recently and it has become necessary to use the technology in the education area. As a part of it, the application of facial recognition technology plays an important role in building the intelligent education system, where teaching and learning can be measured to support quality improvement. The required-attendance taking activity at the schools is one of the important parts of the teaching process. The traditional attendance method causes several inconveniences such as wasting time and effort, easy to mistake and difficult to verify. In this paper, we introduce an automatic attendance system based on the combination of facial recognition technology and interaction with the existing academic portal. We conducted a review of several modern methods to select the most suitable open framework for individual tasks. Our proposed design is flexible and be able to apply to the large-scale set of students without compromising predictive accuracy. We have tested the system on more than 2200 freshmen at FPT University in Hanoi, Vietnam. Initial numerical results show the effectiveness in both aspects of the accuracy and performance of the prototype.