<![CDATA[Student attendance is essential in academic activities, yet conventional methods are prone to fraud, errors, and inefficiency. This study develops an Android-based attendance system integrating face recognition using FaceNet and GPS-based location validation within a client–server architecture. The system performs facial embedding extraction and identity verification using Cosine Similarity with a threshold of 0.7, while geofencing ensures valid attendance locations. Testing on 20 students shows an accuracy of 87.14%, precision of 100%, recall of 83%, and an F-score of 90.71%. GPS validation achieves 100% accuracy within a 10-meter radius. Robustness testing demonstrates stable performance under normal conditions and facial expression variations, with decreased performance under occlusions, extreme lighting, and motion. The system achieves an average processing time of 38.12 ms, indicating fast and responsive performance. These results show that the proposed system improves the accuracy, security, efficiency, and robustness of student attendance management.]]>
Copyrights © 2026
