This paper presents the design and implementation of a novel and robust automated attendance system using computer vision and deep learning techniques. The system aims to modernize attendance management in educational institutions by automating the process of tracking students’ attendance in real time. The proposed system captures multiple images during a class session, recognizing students’ faces to mark attendance, detect latecomers, and ensure students remain in class. We utilize the YOLOv8 model for object detection, where persons are localized in the classroom by generating bounding boxes around them. Then, the bounding boxes are preprocessed and fed to the InsightFace model for recognition. The system captures the first image for recording present students, the second for identifying late students, and a random third image to ensure no students have left the class. If a student is not recognized in the first and second images, they are considered absent. The captured images and attendance results are stored and updated in real time, with a user-friendly interface allowing instructors to manage courses, start attendance sessions, and manually modify attendance if needed. The system provides comprehensive session logs, which can be exported as Excel reports for administrative review. Our experiments carried out in actual classroom environments demonstrate the system’s ability to reduce manual workload, increase accuracy, and enhance the overall classroom management experience. Compared to existing solutions like card-based systems or face recognition technologies such as FaceNet and DeepFace, our system offers unique advantages in real-time tracking and ensuring students remain present throughout the whole session. The results indicate that our automated system is highly scalable and can be integrated with cloud-based services for further enhancement, making it a valuable tool for modern education. Throughout extensive experiments, our system achieved an overall accuracy of 98.51% under various conditions.