Virtual Booth

In recent years, face recognition technology has gained traction across various sectors due to its touchless and non-intrusive nature, despite being slightly less accurate than iris or fingerprint recognition. Traditional attendance methods, such as roll calls, signature sheets, or card punching, are inefficient, time-consuming, and vulnerable to proxy fraud. To address these challenges, this project presents a smart attendance system powered by real-time face recognition. The system uses MTCNN (Multi-task Cascaded Convolutional Networks) for robust face detection and the InceptionResnetV1 model pretrained on the VGGFace2 dataset to extract facial embeddings. These are matched against a pre-registered database using cosine similarity for identity verification. Implemented with Python, OpenCV, and PyTorch, the system processes live video streams to recognize students and log attendance automatically. Experimental results show reliable performance under varying lighting conditions, reducing classroom disruptions while ensuring accurate, contactless, and fraud-resistant attendance tracking. This approach demonstrates the practical potential of face recognition in modernizing attendance management within educational institutions.