Article Per Year (5 Year)
p-Index From 2019 - 2024
0.23
P-Index
This Author published in this journals
All Journal Journal of Data Science and Software Engineering
Ando Hamonangan Saragih
Unknown Affiliation
Computer Science & IT

Published : 1 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 1 Documents
Search

 1 
Implementasi Algoritma Convolutional Neural Network (CNN) Untuk Klasifikasi Gambar X-Ray Penyakit Covid-19 dan Pneumonia Fitria Agustina fitria; Andi Farmadi; Dwi Kartini; Dodon Turianto Nugrahadi; Ando Hamonangan Saragih
Journal of Data Science and Software Engineering Vol 3 No 01 (2022)
Publisher : Fakultas MIPA Universitas Lambung Mangkurat

Show Abstract | Download Original | Original Source | Check in Google Scholar | Full PDF (1494.934 KB)

Abstract

Abstrak Pneumonia caused by the corona virus is different from ordinary pneumonia. One way to find out which pneumonia is caused by the corona virus is to do an X-ray. The disadvantage of this examination is that it requires a radiologist and the analysis time is relatively long. Therefore, to overcome this problem, deep learning methods can be used by implementing the Convolutional Neural Network (CNN) Algorithm method for X-ray image classification. The implementation of the Convolutional Neural Network (CNN) Algorithm is done by using training data of 4800 images which are trained using batch size values ​​of 16, 32, and 64. The train process with batch size values ​​of 16, 32 and 64 produces an average accuracy of 90%, 91% and 92%, while the loss values ​​are 0.22, 0.16 and 0.25. From this process it was found that batch 64 was the best loss and accuracy result for training data. The test data with batch values ​​of 16, 32, and 64 resulted in an accuracy of 76%, 82% and 76%, while the loss values ​​were 0.79, 0.53 and 0.63. The results of this manual testing of 30 photos contained 7 images that are not recognized by the model because of the images look similar to each other with an accuracy of 76%. From this process it was found that batch 32 was the best loss and accuracy result for testing data.
Co-Authors Andi Farmadi Dodon Turianto Nugrahadi Dwi Kartini Fitria Agustina fitria