Automatic Attendance System

This paper presents the design and implementation of an Automatic Attendance System that addresses the limitations of traditional attendance management, such as inefficiency, susceptibility to human error, and lack of data integrity. The proposed system, developed as a native Android application, leverages computer vision and deep learning techniques to ensure accuracy, scalability, and real-time performance. It employs the YOLO (You Only Look Once) algorithm for robust multi-face detection, integrated with Deep Convolutional Neural Networks (DCNN) for reliable facial feature extraction and recognition. Pre-processed biometric data undergoes advanced image transformations to capture both spatial and temporal facial variations, while an ensemble fusion mechanism aggregates outputs from multiple models to minimize false positives and enhance reliability. The Android-based interface enables seamless real-time attendance monitoring, instant visualization of recognition results, and automated generation of attendance reports. The modular architecture of the system comprises key stages: image acquisition, pre-processing, feature extraction, classification, and result reporting. Experimental evaluations demonstrate high accuracy, achieving up to 99.9% recognition with balanced precision-recall metrics, even under varying environmental and attendance conditions. By combining biometric security standards with scalable mobile deployment, this system provides an efficient, accurate, and practical solution for modernizing attendance management in educational institutions, establishing a strong foundation for future enhancements in intelligent classroom systems.