Perbandingan Akurasi Arsitektur Cnn Model Inceptionv3, Mobilenetv2, Dan Vgg16 Pada Sistem Deteksi Pengguna Masker Wajah

Johanes, Christianto Tiku (2022) Perbandingan Akurasi Arsitektur Cnn Model Inceptionv3, Mobilenetv2, Dan Vgg16 Pada Sistem Deteksi Pengguna Masker Wajah. Undergraduate Thesis thesis, Institut Teknologi Telkom Purwokerto.

	Text COVER.pdf Download (347kB)
	Text ABSTRACT.pdf Download (25kB)
	Text ABSTRAK.pdf Download (24kB)
	Text BAB I.pdf Download (151kB)
	Text BAB II.pdf Download (844kB)
	Text BAB III.pdf Download (311kB)
	Text BAB IV.pdf Restricted to Registered users only Download (586kB) | Request a copy
	Text BAB V.pdf Download (26kB)
	Text DAFTAR PUSTAKA.pdf Download (104kB)
	Text LAMPIRAN.pdf Restricted to Registered users only Download (172kB) | Request a copy

Abstract

In 2020, Indonesia has been infected with the Covid 19 virus. The very significant spread of it has forced the government to issue regulations regarding the application of the use of masks to prevent the development of the COVID-19 virus in the community. The problem with the Covid task force officers in monitoring people who wear masks and do not wear masks. By utilizing technological developments in the fields of deep learning and computer vision, this study aims to detect people who wear masks and do not wear masks on their faces. The detection area is around the face. The method used is a convotiunal neural network (CNN). This study compares 3 CNN architecture models, namely MobileNetV2, INCEPTIONV3 and VGG16. The results of the evaluation of the model using the confusion matrix showed the accuracy of INCEPTIONV3 31.65%, MobileV2 33.94%, and VGG16 71.5%. Highest accuracy on VGG16 with batch size 2, epochs 20, validation split 0.2. Suggestions for further research are to add more varied datasets such as images from the left and right sides, images of people wearing masks but blocked by objects and images that lack light. Keywords : covid19, deeplearning, CNN, Open Cv, MobileNetV2, VGG16, Inception V3

Actions (login required)

View Item