Open AccessAnalysis Of Deep Learning Architecture In Classifying SNI Masks
Author(s) -
Fauzan Nurahmadi,
Fahrurrozi Lubis,
Pauzi Ibrahim Nainggolan
Publication year - 2022
Publication title -
jite (journal of informatics and telecommunication engineering)
Language(s) - English
Resource type - Journals
eISSN - 2549-6255
pISSN - 2549-6247
DOI - 10.31289/jite.v5i2.6341
Subject(s) - government (linguistics) , computer science , architecture , computer security , distancing , covid-19 , agency (philosophy) , deep learning , artificial intelligence , medicine , geography , philosophy , linguistics , disease , archaeology , epistemology , pathology , infectious disease (medical specialty)
In preparing for the new normal for COVID-19, every government agency, school and university will be required to comply with new regulations by the government, in which the government will oblige everyone who does activities outside the home to wear masks and practice physical distancing. This is one of the new habits that the government will familiarize with starting in 2020. Due to the ease with which the Covid-19 virus spreads. So the selection of a good mask is recommended good mask, namely a mask that follows the recommendations of the WHO at least 3 layers. The purpose of this study was to classify the types of SNI and non-SNI masks so that the presence of this SNI mask cluster monitoring system could improve security at locations that apply the use of masks and the masks used can function effectively to prevent the spread and spread of Covid-19, classification of research models it uses the InceptionV3, Resnet50, InceptionV2, AlexNet and DenseNet architectures. The results of trials that have been carried out by the InceptionV3 architecture have the most optimal accuracy with loss values of 3.4889 and 0.9894 (98.94%).