DETEKSI CEPAT SUSPECT COVID-19 MENGGUNAKAN DEEP LEARNING DENGAN MEMBANDINGKAN LUNG CT SCAN IMAGES DATASET

Abstract

Abstrak

Dalam tulisan kali ini peneliti akan membahas mengenai pemanfatan cabang keilmuan Kecerdasan Buatan (Artificial Intelegence) yakni pembelajaran dalam (Deep Learning) untuk mengidentifikasi obyek visual Lung CT Scan Images Dataset suspect COVID-19 dengan membandingkannya antara kondisi sehat dan kondisi kritis selama masa inkubasi virus ini. Menggunakan lung CT scan images dataset dan library deep learning tensorflow dengan metode preprocessing image Convolutional Neural Netwok (CNN) diperoleh hasil olah dataset yang berjumlah 4 menggunakan variable epoch 0-4 grafik antara learning loss dan learning accuracy saling berseberangan, dengan epoch semakin besar maka learning loss semakin kecil (variable antara 1,5 - 0,85), berbanding terbalik dengan learning accuracy yang angkanya semakin besar (variable antara 0,44 - 0,64) jika epochnya diperbesar, artinya kemungkinan system akan semakin akurat dengan jumlah training dan dataset yang semakin banyak atau besar.

Kata kunci: Deep Learning, Lung CT Scan Images Dataset, Convolutional Neural Network, Library Tensorflow.

Abstract

In this paper, the researcher will discuss the use of the scientific branch of Artificial Intelligence (Artificial Intelligence), namely deep learning to identify visual objects. CT Scan Images of Lungs Dataset of suspected COVID-19 data by comparing them between healthy and critical conditions during the incubation period. this virus. Using the lung CT scan images dataset and the deep learning tensorflow library with the Convolutional Neural Netwok (CNN) image preprocessing method, it is obtained the results of the data processing which can use 4 epoch variables 0-4 graphs between learning loss and learning accuracy are opposite, with the greater the epoch, the learning loss is getting smaller (variable between 1.5 - 0.85), inversely proportional to the accuracy of learning where the number is getting bigger (variable between 0.44 - 0.64) if the epoch is enlarged, meaning that the system is likely to be more accurate with the amount of training and the dataset which is more or greater.

Keywords: Deep Learning, Lung CT Scan Images Dataset, Convolutional Neural Network, Tensorflow Library