Design of Automated Smart Attendance System Using Deep Learning Based Face Recognition
Ritonga Mahyudin, Domenic T. Sanchez, Sunkara Teena Mrudula, Ravi Kishore Veluri, Jawarneh Malik, Abhishek Raghuvanshi · 2024
The process of taking attendance in the traditional manner is one that is both laborious and time-consuming. Face recognition and identification technology was developed with the primary objective of providing a timesaving automated solution for tracking attendance. This article presents Design of automated smart attendance system using deep learning based face recognition. Camera IoT devices are used to acquire images. These images are stored in cloud via IoT gateway. Images contain many noises. To remove or reduce these noises, adaptive median filter are used. Once noise is removed, then images quality is enhanced by particle swarm optimization. Enhanced images are classified by CNN, CNN VGG 16 and Xception CNN deep learning techniques. Performance is compared on the basis of metrics like- accuracy, specificity and recall. Xception CNN is outperforming other techniques used in the framework. Accuracy, Specificity and Recall rate of Xception CNN is 99.33%, 98.66% and 99.33 percent respectively.