Study on performance of face recognition using convolutional neural network

Rohaizi, Muhammad Aiman (2021) Study on performance of face recognition using convolutional neural network. Project Report. Universiti Teknikal Malaysia Melaka, Melaka, Malaysia. (Submitted)

Text (Full Text) Study on performance of face recognition using convolutional neural network.pdf - Submitted Version Download (5MB)

Abstract

Face recognition has been known worldwide, with daily usage and can be found everywhere. Every person has different features on their face is the reason face recognition is viable. Covid-19 has change how people live in daily life. People need to wear facemask every time they step out of their house. Wearing a facemask can limit the features on the face for face recognition. Thus, in this paper, Convolutional Neural Network (CNN) architecture will be developed and evaluate the performance of the developed CNN in making correct identification of the different type of faces including face that is wearing a facemask. The process will involve setting up datasets, which consist of five categories with four types of face, determining the best settings in CNN and analyzing the results. The results of five different tests were recorded and the final CNN architecture is able to obtain up to 96.2% in validation accuracy. This study shows the performance of CNN in making classification of people wearing facemask is highly accurate.

Actions (login required)

View Item