Facial Recognition based Attendance System Using CNN and Raspberry Pi
Muhammad Owais, Aireen Amir Jalal, Muhammad Moiz Hassan, Ammara Shaikh · 2020 4th International Symposium on Multidisciplinary Studies and Innovative Technologies (ISMSIT) · 2020
This work aims to present face recognition solution using deep learning based facial recognition algorithm. Convolutional Neural Network (CNN) along with triplet loss function has been used to tweak the neural network weights in a way to make the vectors closer via distance metric. The 128-d embedding of each image constitute the feature vectors while K-Nearest Neighbors (KNN) model classifier has been used along with maximum vote count to classify face images. A varying dataset which has been used in this work is the DSU Dataset. The DSU Dataset has been generated locally at DHA Suffa University, Karachi, Pakistan. The implemented algorithm is developed on Python and ensures an overall efficiency of around ninety five percent which is then implemented on Raspberry pi hardware along with an addition of digital attendance management through email.