Article Per Year (5 Year)
p-Index From 2020 - 2025
0.23
P-Index
This Author published in this journals
All Journal Jurnal Pengembangan Teknologi Informasi dan Ilmu Komputer
Javier Ardra Figo
Fakultas Ilmu Komputer, Universitas Brawijaya
Computer Science & IT Control & Systems Engineering Education Electrical & Electronics Engineering Engineering

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

Found 1 Documents
Search

 1 
Deteksi Covid-19 dari Citra X-ray menggunakan Vision Transformer Javier Ardra Figo; Novanto Yudistira; Agus Wahyu Widodo
Jurnal Pengembangan Teknologi Informasi dan Ilmu Komputer Vol 7 No 3 (2023): Maret 2023
Publisher : Fakultas Ilmu Komputer (FILKOM), Universitas Brawijaya

Show Abstract | Download Original | Original Source | Check in Google Scholar

Abstract

Corona Virus is a single stranded RNA virus that can infect human dan a few animal. X-ray Imaging can be one of the few way to check or monitor lungs condition such in the case of tuberculosis, pneumonia, and hernia. Combining X-ray Imaging with deep learning can be one of the solution to the covid-19 detection problem. Vision Tranformer is an architecture that inspired by transformer which is state of the art in the natural language processing realm. One of the few public dataset that contain x-ray image is covidX. CovidX can be breakdown into 3 classes which is pneumonia, covid-19, and normal with as few as 30,530 x-ray image available.the Dataset will processed with data augmentation gaussian blur and colorjitter. The vision transformer that will be used in this experiment is base, large, and huge. This architecture will be used with transfer learning and data augmentation. This experiment will use 40 Epoch, stochastic gradient descent Optimizer, WarmupCosine Scheduler, and Cross Entropy loss function. This experiment will test the effect of transfer learning toward accuracy, the effect of data augmentation toward accuracy, and then will be compared to other architecture. The best accuracy from this experiment is achieved by ResNet50 with transfer learning that achieve accuracy as high as 0.9617006 with validation data and 0.9548872 in test data. Based on this result, the model is overfitting.
Co-Authors Agus Wahyu Widodo Novanto Yudistira