Here is a Comprehensive, IEEE-Style Conference Paper Based on The Project Information Provided

In modern educational institutions, maintaining accurate student attendance is a prerequisite for academic evaluation. Traditional methods, such as manual roll calls or paper-based signing sheets, are time-consuming, prone to human error, and susceptible to manipulation via proxy attendance. This paper proposes an automated Smart Attendance System utilizing computer vision and deep learning techniques. The system employs Multi-task Cascaded Convolutional Networks (MTCNN) for robust face detection and FaceNet for generating high-quality 128-dimensional face embeddings. By capturing video feeds from classroom cameras, the system identifies students in real-time and updates a MySQL/SQLite database automatically. Experimental results on a custom dataset demonstrate that the proposed system achieves high recognition accuracy under varying lighting conditions and head poses, significantly reducing administrative burden and eliminating fraudulent attendance.