Lung Opacity and COVID-19 Chest Radiography Classification and Detection by FasterRCNN with EfficientNet Backbone

Participants

• Dongha Kim - Yonsei Univ. College of Medicine.
• Junho Lee - Yonsei Univ. Dept. of Computer Engineering.

---

Description

X-ray photographs of representative lung diseases, Lung Opacity and COVID-19, will be classified as deep learning, and patient disease will be classified with minimal Cost by detecting each disease site. Correlation analysis of diseases according to gender age and correlation analysis and visualization using various worldwide COVID-19 statistical indicators will be carried out.

---

Datasets

RSNA Pneumonia Detection Challenge Dataset

• https://www.kaggle.com/c/rsna-pneumonia-detection-challenge/overview

COVID-19 Radiography Datset

• https://www.kaggle.com/preetviradiya/covid19-radiography-dataset

---

Model Architecture

EfficientNet

• https://arxiv.org/abs/1905.11946

Faster-RCNN

• https://arxiv.org/abs/1506.01497

---

Source Code

Train Script

• See: train.py

Full Source Code

• See: https://github.com/kdha0727/lung-opacity-and-covid-chest-x-ray-detection

All non-necessary codes are modularized as package. Watch all codes in github repository.

---