Deep Learning based HOG ResNet Model for Group Attendance System in Educational Institutions

Traditional attendance management systems in educational institutions use manual methods that are time-consuming and prone to errors. This work proposes an automated attendance taking system implemented through group pictures using face recognition making the process much easier in classroom environments. For detection of faces from group photographs, the system employs Histogram of Oriented Gradients (HOG) available in the dlib library. For face recognition, Deep learning architecture known as ResNet is used to identify the students under different poses and lighting conditions. A real-time dataset was created using various photographs of the students from different angles with the most common scenarios that are likely to happen in the real classroom situation. The facial recognition model was trained on this custom real-time dataset thus addressing the challenge of photography in the classroom conditions. Using group images, the system recognizes students present classes, marks the attendance and registers them in the attendance database in real time. A user-friendly interface was developed for efficient management of attendance records. Developed Group picture attendance system demonstrates high accuracy in attendance management in various conditions, featuring seamless database integration and a user-friendly front-end interface for data retrieval.