Automatic Face Detection and Recognition for Attendance Maintenance
Narayana Darapaneni, Aruna Kumari Evoori, Vijaya Babu Vemuri, Thangaselvi Arichandrapandian, G. Karthikeyan, Anwesh Reddy Paduri, Dhivakar Babu, J Madhavan
Abstract
This paper focuses on building a deep learning based efficient attendance capturing system. Contemporary world is heading towards AI where every second creates a new vision with an enormous change. In Artificial Intelligence (AI), face recognition is one of the fastest growing domains. Instead of using traditional methods for marking attendance, we propose to automate it by identifying human faces with their unique face features known as Face Recognition. Face detection is a prerequisite process for face recognition which aims to identify and locate all faces irrespective of their position, scale, orientation, lighting conditions, expression etc. We created a system architectural solution using YOLO, MTCNN, FaceNet embeddings by applying multiple augmentations, picture quality check and de-noise methods to get a better attendance system with less maintenance, low cost hardware (Google Colab - Free Version), better performance and accuracy.