-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathtrain_mri.py
More file actions
159 lines (137 loc) · 7.05 KB
/
train_mri.py
File metadata and controls
159 lines (137 loc) · 7.05 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
import torch
import torch.distributed as dist
import torch.multiprocessing as mp
from torch.nn.parallel import DistributedDataParallel as DDP
import torch.nn as nn
import torch.optim as optim
from torch.utils.data import DataLoader
from torch.optim.lr_scheduler import OneCycleLR
import argparse
import wandb
import os
import numpy as np
import random
from tqdm import tqdm
from dipy.io.image import save_nifti
from dataset.dataset_mri import MRIDataset
from utils.utils_mri import AugmentNoise
from network.mri.model import Model
def calculate_psnr(target, ref, data_range=255.0):
img1 = np.array(target, dtype=np.float32)
img2 = np.array(ref, dtype=np.float32)
diff = img1 - img2
psnr = 10.0 * np.log10(data_range**2 / np.mean(np.square(diff)))
return psnr
def train(model, device, train_loader, optimizer, scheduler, epoch, noise_adder, loss_fn):
model.train()
train_loss = 0
progress_bar = tqdm(enumerate(train_loader), total=len(train_loader), desc=f"Epoch {epoch}")
for i, train_data in progress_bar:
train_data['X'] = train_data['X'].to(device)
image = train_data['X']
noise = noise_adder.get_train_noise(image)
pred = model(image, noise)
optimizer.zero_grad()
loss = loss_fn(pred, noise)
train_loss += loss.item()
loss.backward()
optimizer.step()
scheduler.step()
if i % 10 == 0:
progress_bar.set_postfix(loss=loss.item(), loss_avg=train_loss / (i + 1), lr=scheduler.get_last_lr()[0])
wandb.log({"loss": loss.item(), "epoch": epoch, "batch_idx": i, "lr": scheduler.get_last_lr()[0]})
wandb.log({"train_loss": train_loss / (i + 1)})
train_loss /= len(train_loader)
print(f'\nTrain set: Average loss: {train_loss:.4f}')
wandb.log({"train_loss": train_loss})
return loss
def evaluate(model, device, val_loader, noise_adder, loss_fn):
model.eval()
val_loss = 0
progress_bar = tqdm(enumerate(val_loader), total=len(val_loader), desc="Validation")
with torch.no_grad():
for i, val_data in progress_bar:
val_data['X'] = val_data['X'].to(device)
image = val_data['X']
noise = noise_adder.get_train_noise(image)
pred = model(image, noise)
img_loss = loss_fn(pred, noise)
val_loss += img_loss.item()
progress_bar.set_postfix(loss=val_loss / (i + 1))
wandb.log({"val_loss": val_loss / (i + 1), "batch_idx": i})
val_loss /= len(val_loader)
print(f'\nValidation set: Average loss: {val_loss:.4f}\n')
wandb.log({"val_loss": val_loss})
return val_loss
def main_worker(rank, world_size, args):
torch.cuda.set_device(rank)
device = torch.device(f'cuda:{rank}')
dist.init_process_group(backend='nccl', init_method='env://', world_size=world_size, rank=rank)
result_dir = 'results'
os.makedirs(result_dir, exist_ok=True)
wandb.init(project="mri", config=args, name=f"{args.alias}")
torch.manual_seed(args.seed + rank)
np.random.seed(args.seed)
random.seed(args.seed)
torch.cuda.manual_seed(args.seed)
torch.cuda.manual_seed_all(args.seed)
train_dataset = MRIDataset(dataroot=args.dataroot, phase='train', padding=3, in_channel=1)
train_sampler = torch.utils.data.distributed.DistributedSampler(train_dataset, num_replicas=world_size, rank=rank)
train_loader = DataLoader(dataset=train_dataset, sampler=train_sampler, num_workers=0, batch_size=args.batch_size, shuffle=False, pin_memory=True, drop_last=True)
val_dataset = MRIDataset(dataroot=args.dataroot, phase='val', padding=3, in_channel=1)
val_loader = DataLoader(dataset=val_dataset, batch_size=args.test_batch_size, pin_memory=True, shuffle=False)
noise_adder = AugmentNoise(style="gauss45")
model = Model(timesteps=args.timesteps, in_channels=1, base_dim=args.model_base_dim, dim_mults=[2, 4])
model = model.to(device)
model = DDP(model, device_ids=[rank], find_unused_parameters=True)
optimizer = optim.AdamW(model.parameters(), lr=args.lr)
scheduler = OneCycleLR(optimizer, args.lr, total_steps=args.epochs * len(train_loader), pct_start=0.25, anneal_strategy='cos')
loss_fn = nn.MSELoss(reduction='mean')
loss = 0.0
start_epoch = 1
if args.resume != '' and args.resume is not None:
print(f"Resuming training from {args.resume}")
checkpoint = torch.load(args.resume, map_location=device)
model.load_state_dict(checkpoint['model_state_dict'])
optimizer.load_state_dict(checkpoint['optimizer_state_dict'])
start_epoch = checkpoint['epoch'] + 1
loss = checkpoint['loss']
if args.ckpt != '' and args.ckpt is not None:
print(f"loading model from {args.ckpt}")
checkpoint = torch.load(args.ckpt, map_location=device)
model.load_state_dict(checkpoint['model_state_dict'])
for epoch in range(start_epoch, args.epochs + 1):
train_sampler.set_epoch(epoch)
loss = train(model, device, train_loader, optimizer, scheduler, epoch, noise_adder, loss_fn)
if rank == 0:
torch.save({
'epoch': epoch,
'model_state_dict': model.state_dict(),
'optimizer_state_dict': optimizer.state_dict(),
'loss': loss,
}, os.path.join(result_dir, 'model_checkpoint.pth'))
evaluate(model, device, val_loader, noise_adder, loss_fn)
dist.destroy_process_group()
def main():
parser = argparse.ArgumentParser(description='PyTorch Training Script')
parser.add_argument('--batch-size', type=int, default=12, help='input batch size for training (default: 64)')
parser.add_argument('--test-batch-size', type=int, default=8, help='input batch size for testing (default: 1000)')
parser.add_argument('--epochs', type=int, default=100, help='number of epochs to train (default: 10)')
parser.add_argument('--lr', type=float, default=0.0001, help='learning rate (default: 0.01)')
parser.add_argument('--seed', type=int, default=0, help='random seed (default: 1)')
parser.add_argument('--resume', type=str, default='', help='resume training from checkpoint')
parser.add_argument('--ckpt', type=str, default='', help='resume training from checkpoint')
parser.add_argument('--gpus', type=str, default='0', help='Comma separated list of GPU ids to use (default: 0)')
parser.add_argument('--alias', type=str, default='experiment', help='alias for the project')
parser.add_argument('--image_size', type=int, default=256, help='size of the input image')
parser.add_argument('--timesteps', type=int, default=1000, help='iterations for model')
parser.add_argument('--model_base_dim', type=int, default=64, help='base dim of Unet')
parser.add_argument('--dataroot', type=str, default='', help='path to dataset')
args = parser.parse_args()
os.environ['CUDA_VISIBLE_DEVICES'] = args.gpus
world_size = len(args.gpus.split(','))
os.environ['MASTER_ADDR'] = 'localhost'
os.environ['MASTER_PORT'] = '12355'
mp.spawn(main_worker, nprocs=world_size, args=(world_size, args))
if __name__ == '__main__':
main()