Designing an integrated AIOT system for tracking class attendance

Kuak, Xuan Ren (2024) Designing an integrated AIOT system for tracking class attendance. Final Year Project, UTAR.

Preview

PDF Download (2883Kb) | Preview

Abstract

The project aims to develop an automatic facial recognition system that supports artificial intelligence (AI) and the Internet of Things (IoT) for attendance system. The system is able to capture the students' real facial feature data, and uses it as a tool to achieve high-accuracy student identification. Basics and ensures that the software is more secure than the previous traditional attendance method using roll-less slides. For face detection the system uses a YOLO (You Only Look Once) algorithm, which allows quick and efficient recognition of student faces in the classroom context. For facial recognition, deep metric learning methods which involve face encoding are used, and the system can highly match student faces with relevant confidence level of 0.75 or above. In the current work, face recognition is carried out using the ResNet-34 deep convolutional neural network to produce a 128 -dimensional face vectors for the identification. Records for attendance control are kept in the Excel, and this cuts down the time taken to record attendance since Excel has inbuilt facilities for calculating the percentage attendance of each individual. The system shows optimal performance as well as scalability with the overall achievement of the goals that include automation, accuracy and efficiency in attendance tracking. Further enhancements including integrating of night vision cameras and research on face recognition algorithms that utilizes GPU can improve the system performance. The objective of this project is accomplished in the creation of a reliable, scalable and user-friendly system for facial recognition attendance

Actions (login required)