Student Attendance Collection using Deep Learning for Face Detection in Artificial Intelligence

This project proposes a comprehensive face detection and recognition system based on the YOLOv8 model combined with the FaceNet architecture. Strong solutions that can provide real-time identification with advanced detection and recognition of faces are required in the growth of automation in identification technologies within security, healthcare, and retail sectors. Hence, the system proposed depends on the capability of YOLOv8 to detect faces efficiently in one pass, thereby achieving high precision with such speed. Once detected, face detections are used as FaceNet in the generation of facial embeddings to become unique identifiers for a person. The project included real time dataset customized for the problem at hand from Students, training of a YOLOv8 model on this dataset, and then doing inference on test images. The performance of the system was judged by metrices like precision, recall, and mean Average Precision (mAP). Problems under real-world deployments were also mentioned in the paper, which included change in lighting and complete occlusion. It is observed that the system can be able to recognize individuals with reasonable accuracy but, at the same time, be operationally efficient. Future work involves improving the efficiency of algorithms and exploring ethical issues associated with face recognition technology. This work advances ongoing discussion on the uses of face detection and recognition systems in society today.