COVID19-detection is a Deep Learning model to detect infected COVID19 or not from chest X-ray image. ROC AUC is about 96%.
git clone https://github.com/convergence-lab/covid19-detection
clone following dataset repository someware
git clone https://github.com/ieee8023/covid-chestxray-dataset
Change DATADIR_HERE in src/config.toml to the path of the cloned dataset.
[data]
base_dir = "DATADIR_HERE"
Excect train.py
python train.py
胸部X線検査の画像から、コロナウィルスに罹患しているのかどうかを検出するディープラーニングモデルです。 精度は ROC AUC で 96%程度です。
git clone https://github.com/convergence-lab/covid19-detection
どこかに、胸部X線データリポジトリを cloneします
git clone https://github.com/ieee8023/covid-chestxray-dataset
src/config.tomlの DATADIR_HEREの部分に、先ほどクローンしたリポジトリへのパスを書きます。
[data]
base_dir = "DATADIR_HERE"
train.pyを実行します。
python train.py
COVID-19 image data collection
Joseph Paul Cohen and Paul Morrison and Lan Dao COVID-19 image data collection, arXiv:2003.11597, 2020 https://github.com/ieee8023/covid-chestxray-dataset
[I 200424 21:16:31 train:92] Epoch 200 Train loss 0.56104, Acc 74.3%, F1 66.3%, AUC 75.95%, grad 0.9589924216270447
[I 200424 21:16:33 train:112] Epoch 200 Test loss 0.25031, Acc 86.0%, F1 74.4%, AUC 96.84%