Integrated Smart Attendance Tracker Using YOLOv8 and FaceNet with Spotify ANNOY

The goal of this study is to analyze current methods and propose a smart face-recognition-based attendance monitoring that can be used in any type of institution. Current attendance management systems rely on hardcopy logs which are time-consuming and prone to errors and have a significant ecological footprint. We wish to automate this process to make the attendance system more transparent and less prone to errors and introduce a more sustainable approach. There are a number of automated human identification techniques such as biometrics, RFID, eye tracking, etc. but implementing these methods requires expensive hardware systems and resources which can be hard to acquire in smaller institutions. This project proposes a novel face recognition method using YOLOv8 for identifying face bounding boxes, FaceNet for calculating face embeddings, and lastly, Spotify ANNOY equating faces. The proposed method has the major advantage of not requiring any additional hardware components, along with using less computation power; this allows the model to follow sustainable development goals. For comparison, four models were tested, and out of the four, the proposed system performed exceptionally well getting an accuracy of about 95% on a custom dataset.