The development of the Smart Classroom system began with understanding the inefficiencies in traditional attendance methods. We started by researching available face detection techniques, collecting classroom image data, and testing various conditions like lighting and face angles. After selecting OpenCV and the face_recognition library for their performance and ease of use, we designed the architecture to follow a clear flow: capturing video input, detecting faces, matching them against known encodings, and logging attendance

Once the core logic was finalized, we implemented a real time detection and recognition loop using Python. The system encoded student faces from a training set and matched incoming video frames to identify and log attendance automatically. To handle edge cases like unknown faces or missed matches, we integrated alert triggers and fallback logging mechanisms. We also added a visual interface with labeled bounding boxes and a simple terminal log for feedback.

Testing was done across multiple sessions using both recorded and live feeds. We validated the system’s accuracy, tuned the detection thresholds, and ensured the face recognition remained reliable even under varied lighting conditions and slight angle shifts. The final system was lightweight, fast, and accurate, with logs saved to CSV for potential dashboard use later.