AIMI-lab1

Training model to detect Pneumonia from chest X-ray images

This project uses Pytorch to train model. You must check you have pytorch and GPU for training.

Train

1. Download train.py
2. Check your environment have python, GPU and Pytorch
3. Adjustment parameters for training
4. Start training.

python train.py

Adjustment parameters

#儲存圖片的編號
num = 16  #images name number

#繪製accuracy plot
plot_accuracy(train_acc_list, val_acc_list)

#繪製accuracy plot
plot_f1_score(f1_score_list)

#繪製confusion_matrix
plot_confusion_matrix(confusion_matrix)

#Training基本設定
parser.add_argument('--num_epochs', type=int, required=False, default=35)
parser.add_argument('--batch_size', type=int, required=False, default=32)
parser.add_argument('--lr', type=float, default=1e-5)  #1e-5
parser.add_argument('--wd', type=float, default=0.9)

#Data Augementation
train_dataset = ImageFolder(root=os.path.join(args.dataset, 'train'),
                            transform = transforms.Compose([
                                        transforms.Resize((args.resize, args.resize)),        
                                        transforms.RandomRotation(args.degree, resample=False),
                                        transforms.RandomHorizontalFlip(),
                                        transforms.RandomVerticalFlip(),
                                        transforms.ToTensor(),
                                        ]))  #transforms.Normalize(mean, std)

#Train model
model = models.resnext50_32x4d(pretrained=True)

#Optimizer
optimizer = optim.SGD(model.parameters(), lr=args.lr, weight_decay=args.wd)

Environment

Anaconda env

#更新apt
sudo apt update
#安裝curl
sudo apt install curl
# 下載 Anaconda 安裝檔案
curl -O https://repo.anaconda.com/archive/Anaconda3-2019.10-Linux-x86_64.sh
bash Anaconda3-2019.10-Linux-x86_64.sh

# 生效conda 指令
conda init
source ~/.bashrc
export PATH=~/anaconda3/bin:$PATH

#查看環境
conda info --env
#建立新環境
conda create --name myenv python=3.7
#建立環境錯誤
conda clean -i
conda clean -a

#啟動新環境
source activate myenv

Pytorch

conda install pytorch==1.8.0 torchvision==0.8.0 torchaudio==0.7.0 cudatoolkit=10.1 -c pytorch