AttendAI: Temporal AI for Smarter Class Attendance

AttendAI is an AI-powered classroom attendance system designed to reduce the time and effort involved in manual attendance while making attendance decisions more explainable and robust. Unlike traditional face-recognition systems that mark a student present after detecting them in a single frame, AttendAI uses Temporal Presence Verification. During a lecture, the system captures periodic classroom snapshots, detects and recognizes enrolled students, and records their observations across multiple moments. Attendance is then determined using a configurable minimum-observation rule rather than a single detection. The MVP includes student enrollment, face recognition, periodic snapshot scheduling, configurable observation thresholds, face-quality handling, unknown-face detection, ambiguous recognition handling, attendance timelines, session history, CSV export, and recovery for interrupted sessions. Snapshot timing is distributed across the lecture with controlled jitter rather than following a predictable fixed pattern. AttendAI uses clear observation states such as PRESENT, NEEDS REVIEW, UNKNOWN, LOW QUALITY, and AMBIGUOUS, allowing uncertain cases to remain reviewable instead of forcing an automatic decision. The system is built with Python, OpenCV, a pretrained vision/face-recognition model, SQLite, and Streamlit, with automated tests covering core attendance, scheduling, database, and session logic. Privacy and responsible AI are central considerations because facial data is sensitive biometric information, particularly in educational environments. The system is intended for authorized use with appropriate consent, access controls, data minimization, retention policies, and human oversight. Future development could include head-pose and face-orientation analysis, improved vision models, multi-camera classrooms, authorized Duty/Medical Leave integration, and examination-room workflows, subject to institutional policies, privacy safeguards, and human review.