AnnLIU15/clfCOVID

COVID classification

新冠肺炎分类

Verion2.0 Author: ZhaoY

运行环境

Version	v1.0 20210611
编程语言	Python
Cuda版本	10.0
库	requirements

分割神经网络模型

网络名称	原始文章
ResNet18	Deep Residual Learning for Image Recognition

st=>start: start
rs=>operation: utils/rename_clf_pic
ps=>operation: utils/preprocessClf
fe=>operation: utils/findDismatchSize
de=>operation: utils/delDismatchSizeFile
tr=>operation: clfTrain
ts=>operation: clfTest
e=>end: end

st->rs->ps->fe->de->tr->ts->e

1. 分类(先看分割代码)

预处理分类图片

python ./utils/preprocessClf.py --in_dir /home/e201cv/Desktop/covid_data/clf/train /home/e201cv/Desktop/covid_data/clf/val /home/e201cv/Desktop/covid_data/clf/test --out_dir /home/e201cv/Desktop/covid_data/process_clf/train /home/e201cv/Desktop/covid_data/process_clf/val /home/e201cv/Desktop/covid_data/process_clf/test

获取掩模

python infer.py --infer_data_dirs /home/e201cv/Desktop/covid_data/process_clf/train /home/e201cv/Desktop/covid_data/process_clf/val /home/e201cv/Desktop/covid_data/process_clf/test --pth /home/e201cv/Desktop/covid_seg/output/saved_models/U2Net/bestSeg.pth --num_classes 3 --device cuda

训练1

/home/e201cv/.conda/envs/pt171/bin/python --device cuda --num_classes 3 --radiomics_require False --match False --batch_size 64 --num_epochs 155 --lrate 1e-3 --model_name resnet18

/home/e201cv/.conda/envs/pt171/bin/python clfTrain.py --device cuda --num_classes 3 --radiomics_require False --match False --batch_size 32 --num_epochs 155 --lrate 1e-3 --model_name resnet34

/home/e201cv/.conda/envs/pt171/bin/python clfTrain.py --device cuda --num_classes 3 --radiomics_require False --match False --batch_size 16 --num_epochs 155 --lrate 1e-3 --model_name resnet50

/home/e201cv/.conda/envs/pt171/bin/python clfTrain.py --device cuda --num_classes 3 --radiomics_require False --match False --batch_size 64 --num_epochs 155 --lrate 1e-3 --model_name vgg11

python clfTrain.py --device cuda --num_classes 3 --radiomics_require False --match False --batch_size 12 --num_epochs 80 --lrate 1e-3  --train_data_dir data/process_clf/train/ --val_data_dir data/process_clf/val/ --model_name vgg19

测试

python clfTest.py --num_classes 3 --device cuda --pth output/saved_models/ClfBestModel.pth --batch_size 256 --radiomics_require False --test_data_dir ./data/process_clf/test --save_clf ./output/clfResult/

python clfTest.py --num_classes 3 --device cuda --pth output/saved_models/resnet18/epoch_150_model.pth --batch_size 256 --radiomics_require False --test_data_dir ./data/process_clf/test --save_clf ./output/clfResult/ --model_name resnet18

ROC曲线与具体评测效果显示

python plt_roc.py
python cmp2file.py

log

tensorboard --logdir='log/clf/resnet0607-1902'

外源测试(x)

python extendTest.py --num_classes 3 --device cuda --pth output/saved_models/ClfBestModel.pth --batch_size 1 --test_data_dir ./data/extend --save_clf ./output/clfResult/ --model_name resnet_ex

看学习率

/home/e201cv/.conda/envs/pt171/bin/python /home/e201cv/Desktop/COVID_prj/Code/covid_clf/watchLrate.py -m output/saved_models/resnet/epoch_140_model.pth