import argparse import os import random import warnings import numpy as np import torch import torch.nn as nn import torchvision from PIL import Image from timm.scheduler import CosineLRScheduler from torch.utils.data import ConcatDataset, DataLoader, Dataset from util import (ParamDiffAug, TensorDataset, epoch2, get_daparam, get_dataset, get_network) from utils import DiceLoss from utils.covid19_dataset import STNAugment warnings.filterwarnings("ignore", category=DeprecationWarning) class TestDiceLoss(nn.Module): """Dice loss, need one hot encode input Args: weight: An array of shape [num_classes,] ignore_index: class index to ignore predict: A tensor of shape [N, C, *] target: A tensor of same shape with predict other args pass to BinaryDiceLoss Return: same as BinaryDiceLoss """ def __init__(self, weight=None, ignore_index=None, **kwargs): super(TestDiceLoss, self).__init__() self.kwargs = kwargs self.weight = weight self.ignore_index = ignore_index self.epsilon = 1e-5 def forward(self, predict, target): assert predict.size() == target.size(), "the size of predict and target must be equal." num = predict.size(0) pre = predict.view(num, -1) tar = target.view(num, -1) intersection = (pre * tar).sum(-1) # 利用预测值与标签相乘当作交集 union = (pre + tar).sum(-1) score = 2 * (intersection + self.epsilon) / (union + self.epsilon) return score.mean() class PairDatset(Dataset): def __init__(self, data_path,if_randaugment=False): self.data_path = data_path self.images = [] self.masks = [] self.if_randaugment = if_randaugment self.turn = torchvision.transforms.ToTensor() for root, dirs, files in os.walk(data_path): for file in files: if len(self.images)>500: break path = str(os.path.join(self.data_path, file)) if file.startswith("image_"): self.images.append(path) elif file.startswith("mask_"): self.masks.append(path) else: continue self.indexs = [i for i in range(len(self.images))] self.data_aug = STNAugment() def apply_transforms(self, image, mask, transform, seed=None): if transform is not None: turn_list = [image, mask] turn_list = self.data_aug(turn_list) return turn_list[0], turn_list[1] def __len__(self): return len(self.indexs) def __getitem__(self, item): image_path = os.path.join(self.data_path, "image_" + str(self.indexs[item]) + ".png") mask_path = os.path.join(self.data_path, "mask_" + str(self.indexs[item]) + ".png") image, mask = Image.open(image_path).convert("L"), Image.open(mask_path).convert("L") image, mask = self.turn(image), self.turn(mask) mask = (mask > 0.5).float() if self.if_randaugment: return self.apply_transforms(image, mask, self.data_aug) return image, mask def main(args): with open("./outputs/" + f"{args.generate_data_path.split('/')[-1]}" + f"_model_{args.model}_no_{random.random()}.txt", "w") as ff: # f"{args.generate_data_path.split('/')[-1]}" args.dsa = True if args.dsa == 'True' else False args.device = 'cuda' if torch.cuda.is_available() else 'cpu' args.dsa_param = ParamDiffAug() from utils.cgmh_dataset import split_train_and_val if args.dataset == "COVID19": from utils.covid19_dataset import COVID19Dataset, clean_dataset assert args.csv_path != "no", "COVID-19 Segmentation task need csv metadata!" dst = COVID19Dataset(imgpath=args.data_path, csvpath=args.csv_path, semantic_masks=True) dst = clean_dataset(dst) dst_train, dst_test = split_train_and_val(dst,split_ratio=args.ratio) dst_train_2 = PairDatset(args.generate_data_path) dst_train = ConcatDataset([dst_train_2,dst_train]) elif args.dataset == "CGMH": from utils.cgmh_dataset import CGMHDataset dst_train = CGMHDataset(args.data_path,if_val=True) dst_train, dst_test = split_train_and_val(dst_train) dst_train_2 = PairDatset(args.generate_data_path) dst_train = ConcatDataset([dst_train_2,dst_train]) else: raise NotImplementedError # print('\n================== Exp %d ==================\n '%exp) print('Hyper-parameters: \n', args.__dict__) save_dir = os.path.join("./checkpoint", args.dataset) if not os.path.exists(save_dir): os.makedirs(save_dir) ''' organize the real dataset ''' print("BUILDING DATASET") print("total train images %d" % (len(dst_train))) if args.loss_type == "cross": criterion = nn.CrossEntropyLoss().to(args.device) elif args.loss_type == "l1": criterion = nn.L1Loss().to(args.device) elif args.loss_type == "sigmoid_l1": c_loss = nn.L1Loss().to(args.device) criterion = lambda x, y, tau=1: c_loss(torch.sigmoid(x / tau), y) elif args.loss_type == "bce": c_loss = nn.BCELoss().to(args.device) criterion = lambda x, y: c_loss(torch.sigmoid(x), y) else: raise NotImplementedError criterion_dice = DiceLoss() criterion_test_dice = TestDiceLoss() trainloader = torch.utils.data.DataLoader(dst_train, batch_size=args.batch_train, shuffle=True, num_workers=4) testloader = torch.utils.data.DataLoader(dst_test, batch_size=args.batch_train, shuffle=True, num_workers=4) ''' Train synthetic data ''' teacher_net = get_network(args.model, 1, 1, 256).to(args.device) # get a random model teacher_net.train() lr = args.lr_teacher teacher_optim = torch.optim.Adam(teacher_net.parameters(), lr=lr, weight_decay=args.l2) # optimizer_img for synthetic data teacher_optim.zero_grad() scheduler = CosineLRScheduler(teacher_optim, args.train_epochs * len(trainloader), lr_min=1e-7, warmup_lr_init=lr * 0.01, warmup_t=5 * len(trainloader), t_in_epochs=False) scaler = torch.cuda.amp.GradScaler() iter = 0 for e in range(args.train_epochs): train_loss, train_dice, train_psnr = epoch2("train", dataloader=trainloader, net=teacher_net, optimizer=teacher_optim, scheduler=scheduler, iter=iter, scaler=scaler, criticion=criterion, criticion_dice=criterion_dice, args=args) test_loss, test_dice, test_psnr = epoch2("test", dataloader=testloader, net=teacher_net, optimizer=None, scheduler=scheduler, iter=iter, scaler=scaler, criticion=criterion, criticion_dice=criterion_test_dice, args=args) iter += len(trainloader) log = "Epoch: {}\tIter: {}\tLr: {}\tTrain PSNR: {}\tTrain DICE: {}\tTest PSNR: {}\tTest DICE: {}".format(e, iter, scheduler._get_lr( iter)[ 0], train_psnr, train_dice, test_psnr, test_dice) print(log) ff.write(log + "\n") # # print("Saving {}".format(os.path.join(save_dir, "unet_for_cgmh_fid.pt"))) # torch.save(teacher_net.state_dict(), os.path.join(save_dir, "unet_for_cgmh_fid.pt")) if __name__ == '__main__': parser = argparse.ArgumentParser(description='Parameter Processing') parser.add_argument('--dataset', type=str, default='CGMH', help='dataset') parser.add_argument('--model', type=str, default='Unet', help='model') parser.add_argument('--loss_type', type=str, default='sigmoid_l1', help='loss type') parser.add_argument('--lr_teacher', type=float, default=0.01, help='learning rate for updating network parameters') parser.add_argument('--batch_train', type=int, default=16, help='batch size for training networks') parser.add_argument('--batch_real', type=int, default=16, help='batch size for real loader') parser.add_argument('--dsa', type=str, default='True', choices=['True', 'False'], help='whether to use differentiable Siamese augmentation.') parser.add_argument('--data_path', type=str, default='./CGMH_PelvisSegment', help='dataset path') parser.add_argument('--train_epochs', type=int, default=50) parser.add_argument("--generate_data_path", type=str, default="./output/CGMH/tau_0.5_scale_1.0") parser.add_argument('--zca', action='store_true') parser.add_argument('--ratio', type=float,default=0.9) parser.add_argument('--decay', action='store_true') parser.add_argument('--mom', type=float, default=0, help='momentum') parser.add_argument('--l2', type=float, default=0, help='l2 regularization') parser.add_argument('--save_interval', type=int, default=5) parser.add_argument('--csv_path', type=str, default="no") args = parser.parse_args() main(args) """ python eval.py --dataset=CGMH --loss_type sigmoid_l1 --model=Unet --train_epochs=50 \ --data_path=./CGMH_PelvisSegment \ --csv_path=./covid-chestxray-dataset-master/metadata.csv """