def test():
print('loading test data......')
args = DefaultConfig()
model = generate_model(args)
for dataset in tqdm.tqdm(
args.testdataset,
desc='Total TestSet',
total=len(args.testdataset),
position=0,
bar_format='{desc:<30}{percentage:3.0f}%|{bar:50}{r_bar}'):
dataset_path = os.path.join(args.data, dataset)
dataset_test = TestDataset(dataset_path,
scale=(args.crop_height, args.crop_width),
mode='val')
dataloader_test = DataLoader(dataset_test,
batch_size=1,
shuffle=False,
num_workers=args.num_workers,
pin_memory=True,
drop_last=False)
save_path = os.path.join(args.data, dataset, "output/")
# 判断结果
if not os.path.exists(save_path):
os.makedirs(save_path)
model.eval()
with torch.no_grad():
for i, (img, gt, name) in tqdm.tqdm(
enumerate(dataloader_test),
desc=dataset + ' - Test',
total=len(dataloader_test),
position=1,
leave=False,
bar_format='{desc:<30}{percentage:3.0f}%|{bar:50}{r_bar}'):
if torch.cuda.is_available() and args.use_gpu:
img = img.cuda()
gt = gt.cuda()
output = model(img)
out = F.upsample(output,
size=gt.shape[2:],
mode='bilinear',
align_corners=False)
out = out.data.sigmoid().cpu().numpy().squeeze()
out = (out - out.min()) / (out.max() - out.min() + 1e-8)
#save_path = os.path.join(save_path, name)
path = save_path + "".join(name)
imageio.imwrite(path, out)
def main():
args = DefaultConfig()
"""
create dataset and dataloader
"""
dataset_train = Dataset(args.train_data_path, scale=(args.trainsize, args.trainsize),augmentations=args.augmentations)
dataloader_train = DataLoader(
dataset_train,
batch_size=args.batch_size,
shuffle=True,
num_workers=args.num_workers,
pin_memory=True,
drop_last=True
)
# build model
os.environ['CUDA_VISIBLE_DEVICES'] = args.cuda
"""
load model
"""
model_all={'MyNet':MyNet()}
model=model_all[args.net_work]
print(args.net_work)
cudnn.benchmark = True
# model._initialize_weights()
if torch.cuda.is_available() and args.use_gpu:
model = torch.nn.DataParallel(model).cuda()
"""
optimizer
"""
if args.optimizer == 'AdamW':
print("using AdamW")
optimizer = torch.optim.AdamW(model.parameters(), args.lr,weight_decay=1e-4)
else:
print("using SGD")
optimizer = torch.optim.SGD(model.parameters(), lr=args.lr, momentum=args.momentum, weight_decay=args.weight_decay)
total = len(dataloader_train)
train(args, model, optimizer,dataloader_train,total)
def eval():
args = DefaultConfig()
print('#' * 20, 'Start Evaluation', '#' * 20)
for dataset in tqdm.tqdm(args.testdataset, total=len(args.testdataset), position=0,
bar_format='{desc:<30}{percentage:3.0f}%|{bar:50}{r_bar}'):
pred_path = 'E:\dataset\data\TestDataset/{}/output/'.format(dataset)
gt_path = 'E:\dataset\data\TestDataset/{}/masks/'.format(dataset)
preds = os.listdir(pred_path)
gts = os.listdir(gt_path)
total_batch = len(preds)
# metrics_logger initialization
metrics = Metrics(['recall', 'specificity', 'precision', 'F1', 'F2',
'ACC_overall', 'IoU_poly', 'IoU_bg', 'IoU_mean', 'Dice'])
for i, sample in tqdm.tqdm(enumerate(zip(preds, gts)), desc=dataset + ' - Evaluation', total=len(preds),
position=1, leave=False, bar_format='{desc:<30}{percentage:3.0f}%|{bar:50}{r_bar}'):
pred, gt = sample
assert os.path.splitext(pred)[0] == os.path.splitext(gt)[0]
pred_mask = np.array(Image.open(os.path.join(pred_path, pred)))
gt_mask = np.array(Image.open(os.path.join(gt_path, gt)))
if len(pred_mask.shape) != 2:
pred_mask = pred_mask[:, :, 0]
if len(gt_mask.shape) != 2:
gt_mask = gt_mask[:, :, 0]
assert pred_mask.shape == gt_mask.shape
gt_mask = gt_mask.astype(np.float64) / 255
pred_mask = pred_mask.astype(np.float64) / 255
gt_mask = torch.from_numpy(gt_mask)
pred_mask = torch.from_numpy(pred_mask)
_recall, _specificity, _precision, _F1, _F2, \
_ACC_overall, _IoU_poly, _IoU_bg, _IoU_mean, _Dice = evaluate(pred_mask, gt_mask, 0.5)
metrics.update(recall=_recall, specificity=_specificity, precision=_precision,
F1=_F1, F2=_F2, ACC_overall=_ACC_overall, IoU_poly=_IoU_poly,
IoU_bg=_IoU_bg, IoU_mean=_IoU_mean, Dice=_Dice
)
metrics_result = metrics.mean(total_batch)
print("Test Result:")
print('recall: %.4f, specificity: %.4f, precision: %.4f, F1: %.4f, F2: %.4f, '
'ACC_overall: %.4f, IoU_poly: %.4f, IoU_bg: %.4f, IoU_mean: %.4f, Dice:%.4f'
% (metrics_result['recall'], metrics_result['specificity'], metrics_result['precision'],
metrics_result['F1'], metrics_result['F2'], metrics_result['ACC_overall'],
metrics_result['IoU_poly'], metrics_result['IoU_bg'], metrics_result['IoU_mean'],
metrics_result['Dice']))
from torch.autograd import Variable
import models
from config.config import DefaultConfig
from utils.dataset import TGSSaltDataSet
from utils.utils import *
from utils.metrics import *
from utils.visualize import *
from utils.loss import *
from tqdm import tqdm
import gc
import matplotlib.pyplot as plt
import os
os.environ["CUDA_VISIBLE_DEVICES"] = "1"
opt = DefaultConfig()
import logging
# 通过下面的方式进行简单配置输出方式与日志级别
logging.basicConfig(filename=os.path.join('./result',opt.logging_file), level=logging.INFO)
def test(bestModel, test_dataloader):
if opt.use_tta:
print('use tta...')
test_df['tta_pred_fold%d' %n_fold] = np.nan
test_df['tta_pred_fold%d' %n_fold] = test_df['tta_pred_fold%d' %n_fold].astype(object)
test_unaugment = test_unaugment_flip
else:
print('not use tta...')
test_df['pred_fold%d' %n_fold] = np.nan
import os
import numpy as np
from sklearn.metrics import jaccard_similarity_score
import pandas as pd
from config.config import DefaultConfig
from skimage.transform import resize
from sklearn.model_selection import BaseCrossValidator, train_test_split
from PIL import Image
opt = DefaultConfig()
def GenDataNameDict():
np.random.seed(opt.seed)
DataNameDict = {}
AllTrainNameList = [
trainname.split('.')[-2]
for trainname in os.listdir(os.path.join(opt.train_data_images_root))
]
np.random.shuffle(AllTrainNameList)
AllTestNameList = [
testname.split('.')[-2]
for testname in os.listdir(os.path.join(opt.test_data_root))
]
ValNameList = AllTrainNameList[:int(opt.valPercent *
len(AllTrainNameList))] # 取百分之二十作为验证集
TrainNameList = AllTrainNameList[int(opt.valPercent *
len(AllTrainNameList)):]
DataNameDict['train'] = TrainNameList
DataNameDict['val'] = ValNameList
DataNameDict['test'] = AllTestNameList
def charnge():
args = DefaultConfig()
save_path = os.path.join(args.data, args.dataset, "output")
print(save_path)
arr = []
arr2 = []
for f in os.listdir(destimg):
arr.append(f)
for f in os.listdir(destmask):
if f in arr:
arr2.append(f)
for f in arr:
if f not in arr2:
print(f)
def test8():
for f in os.listdir(
r"E:\dataset\dataset-video\dataset\TestDataset\CVC-ColonDB-300\masks"
):
path = r"E:\dataset\dataset-video\dataset\TestDataset\CVC-ColonDB-300\masks\\" + f
new_name = path.replace(".png", "-CVC-300.png")
os.rename(path, new_name)
from torchvision import transforms
from PIL import Image
if __name__ == '__main__':
from config.config import DefaultConfig
arg = DefaultConfig()
while True:
time.sleep(1)
print(arg.num_epochs)