This study examines the application of the modified Viola-Jones algorithm for student facial recognition at STMIK YMI Tegal, aiming to improve the efficiency and safety of the student attendance system. By adapting the algorithm to address the challenge of facial recognition accuracy from different angles and lighting conditions, a quasi-experimental quantitative design involved collecting data through photographic sessions with student subjects, followed by preprocessing to improve the quality of the analysis. The modification was evaluated for its ability to handle variations in facial and lighting conditions, showing significant improvements with 60% accuracy and precision, recall, and an F1-score of 71.43%. These findings demonstrate the effectiveness of the modification in improving facial recognition, potentially contributing significantly to attendance management and safety practices in educational settings. This research not only strengthens the existing literature.