def train():
device = torch.device(conf.cuda if torch.cuda.is_available() else "cpu")
dataset = Training_Dataset(conf.data_path_train, conf.gaussian_noise_param,
conf.crop_img_size)
dataset_length = len(dataset)
train_loader = DataLoader(dataset,
batch_size=4,
shuffle=True,
num_workers=4)
model = UNet(in_channels=conf.img_channel, out_channels=conf.img_channel)
criterion = nn.MSELoss()
model = model.to(device)
optim = Adam(model.parameters(),
lr=conf.learning_rate,
betas=(0.9, 0.999),
eps=1e-8,
weight_decay=0,
amsgrad=True)
scheduler = lr_scheduler.StepLR(optim, step_size=100, gamma=0.5)
model.train()
print(model)
print("Starting Training Loop...")
since = time.time()
for epoch in range(conf.max_epoch):
print('Epoch {}/{}'.format(epoch, conf.max_epoch - 1))
print('-' * 10)
running_loss = 0.0
scheduler.step()
for batch_idx, (source, target) in enumerate(train_loader):
source = source.to(device)
target = target.to(device)
optim.zero_grad()
denoised_source = model(source)
loss = criterion(denoised_source, target)
loss.backward()
optim.step()
running_loss += loss.item() * source.size(0)
print('Current loss {} and current batch idx {}'.format(
loss.item(), batch_idx))
epoch_loss = running_loss / dataset_length
print('{} Loss: {:.4f}'.format('current ' + str(epoch), epoch_loss))
if (epoch + 1) % conf.save_per_epoch == 0:
save_model(model, epoch + 1)
time_elapsed = time.time() - since
print('Training complete in {:.0f}m {:.0f}s'.format(
time_elapsed // 60, time_elapsed % 60))
out_norm = 15383.0
LE_img = imgread(os.path.join(dir_path_LE, "LE_01.tif"))
LE_512 = cropImage(LE_img, IMG_SHAPE[0],IMG_SHAPE[1])
sample_le = {}
for le_512 in LE_512:
tiles = crop_prepare(le_512, CROP_STEP, IMG_SIZE)
for n,img in enumerate(tiles):
if n not in sample_le:
sample_le[n] = []
img = transform.resize(img,(IMG_SIZE*2, IMG_SIZE*2),preserve_range=True,order=3)
sample_le[n].append(img)
SNR_model = UNet(n_channels=15, n_classes=15)
print("{} paramerters in total".format(sum(x.numel() for x in SNR_model.parameters())))
SNR_model.cuda(cuda)
SNR_model.load_state_dict(torch.load(SNR_model_path))
# SNR_model.load_state_dict(torch.load(os.path.join(dir_path,"model","LE_HE_mito","LE_HE_0825.pkl")))
SNR_model.eval()
SIM_UNET = UNet(n_channels=15, n_classes=1)
print("{} paramerters in total".format(sum(x.numel() for x in SIM_UNET.parameters())))
SIM_UNET.cuda(cuda)
SIM_UNET.load_state_dict(torch.load(SIM_UNET_model_path))
# SIM_UNET.load_state_dict(torch.load(os.path.join(dir_path,"model","HE_HER_mito","HE_X2_HER_0825.pkl")))
SIM_UNET.eval()
SRRFDATASET = ReconsDataset(
img_dict=sample_le,
transform=ToTensor(),
def train(cont=False):
# for tensorboard tracking
logger = get_logger()
logger.info("(1) Initiating Training ... ")
logger.info("Training on device: {}".format(device))
writer = SummaryWriter()
# init model
aux_layers = None
if net == "SETR-PUP":
aux_layers, model = get_SETR_PUP()
elif net == "SETR-MLA":
aux_layers, model = get_SETR_MLA()
elif net == "TransUNet-Base":
model = get_TransUNet_base()
elif net == "TransUNet-Large":
model = get_TransUNet_large()
elif net == "UNet":
model = UNet(CLASS_NUM)
# prepare dataset
cluster_model = get_clustering_model(logger)
train_dataset = CityscapeDataset(img_dir=data_dir,
img_dim=IMG_DIM,
mode="train",
cluster_model=cluster_model)
valid_dataset = CityscapeDataset(img_dir=data_dir,
img_dim=IMG_DIM,
mode="val",
cluster_model=cluster_model)
train_loader = DataLoader(train_dataset,
batch_size=batch_size,
shuffle=True)
valid_loader = DataLoader(valid_dataset,
batch_size=batch_size,
shuffle=False)
logger.info("(2) Dataset Initiated. ")
# optimizer
epochs = epoch_num if epoch_num > 0 else iteration_num // len(
train_loader) + 1
optim = SGD(model.parameters(),
lr=lrate,
momentum=momentum,
weight_decay=wdecay)
# optim = Adam(model.parameters(), lr=lrate)
scheduler = lr_scheduler.MultiStepLR(
optim, milestones=[int(epochs * fine_tune_ratio)], gamma=0.1)
cur_epoch = 0
best_loss = float('inf')
epochs_since_improvement = 0
# for continue training
if cont:
model, optim, cur_epoch, best_loss = load_ckpt_continue_training(
best_ckpt_src, model, optim, logger)
logger.info("Current best loss: {0}".format(best_loss))
with warnings.catch_warnings():
warnings.simplefilter("ignore")
for i in range(cur_epoch):
scheduler.step()
else:
model = nn.DataParallel(model)
model = model.to(device)
logger.info("(3) Model Initiated ... ")
logger.info("Training model: {}".format(net) + ". Training Started.")
# loss
ce_loss = CrossEntropyLoss()
if use_dice_loss:
dice_loss = DiceLoss(CLASS_NUM)
# loop over epochs
iter_count = 0
epoch_bar = tqdm.tqdm(total=epochs,
desc="Epoch",
position=cur_epoch,
leave=True)
logger.info("Total epochs: {0}. Starting from epoch {1}.".format(
epochs, cur_epoch + 1))
for e in range(epochs - cur_epoch):
epoch = e + cur_epoch
# Training.
model.train()
trainLossMeter = LossMeter()
train_batch_bar = tqdm.tqdm(total=len(train_loader),
desc="TrainBatch",
position=0,
leave=True)
for batch_num, (orig_img, mask_img) in enumerate(train_loader):
orig_img, mask_img = orig_img.float().to(
device), mask_img.float().to(device)
if net == "TransUNet-Base" or net == "TransUNet-Large":
pred = model(orig_img)
elif net == "SETR-PUP" or net == "SETR-MLA":
if aux_layers is not None:
pred, _ = model(orig_img)
else:
pred = model(orig_img)
elif net == "UNet":
pred = model(orig_img)
loss_ce = ce_loss(pred, mask_img[:].long())
if use_dice_loss:
loss_dice = dice_loss(pred, mask_img, softmax=True)
loss = 0.5 * (loss_ce + loss_dice)
else:
loss = loss_ce
# Backward Propagation, Update weight and metrics
optim.zero_grad()
loss.backward()
optim.step()
# update learning rate
for param_group in optim.param_groups:
orig_lr = param_group['lr']
param_group['lr'] = orig_lr * (1.0 -
iter_count / iteration_num)**0.9
iter_count += 1
# Update loss
trainLossMeter.update(loss.item())
# print status
if (batch_num + 1) % print_freq == 0:
status = 'Epoch: [{0}][{1}/{2}]\t' \
'Loss {loss.val:.4f} ({loss.avg:.4f})\t'.format(epoch+1, batch_num+1, len(train_loader), loss=trainLossMeter)
logger.info(status)
# log loss to tensorboard
if (batch_num + 1) % tensorboard_freq == 0:
writer.add_scalar(
'Train_Loss_{0}'.format(tensorboard_freq),
trainLossMeter.avg,
epoch * (len(train_loader) / tensorboard_freq) +
(batch_num + 1) / tensorboard_freq)
train_batch_bar.update(1)
writer.add_scalar('Train_Loss_epoch', trainLossMeter.avg, epoch)
# Validation.
model.eval()
validLossMeter = LossMeter()
valid_batch_bar = tqdm.tqdm(total=len(valid_loader),
desc="ValidBatch",
position=0,
leave=True)
with torch.no_grad():
for batch_num, (orig_img, mask_img) in enumerate(valid_loader):
orig_img, mask_img = orig_img.float().to(
device), mask_img.float().to(device)
if net == "TransUNet-Base" or net == "TransUNet-Large":
pred = model(orig_img)
elif net == "SETR-PUP" or net == "SETR-MLA":
if aux_layers is not None:
pred, _ = model(orig_img)
else:
pred = model(orig_img)
elif net == "UNet":
pred = model(orig_img)
loss_ce = ce_loss(pred, mask_img[:].long())
if use_dice_loss:
loss_dice = dice_loss(pred, mask_img, softmax=True)
loss = 0.5 * (loss_ce + loss_dice)
else:
loss = loss_ce
# Update loss
validLossMeter.update(loss.item())
# print status
if (batch_num + 1) % print_freq == 0:
status = 'Validation: [{0}][{1}/{2}]\t' \
'Loss {loss.val:.4f} ({loss.avg:.4f})\t'.format(epoch+1, batch_num+1, len(valid_loader), loss=validLossMeter)
logger.info(status)
# log loss to tensorboard
if (batch_num + 1) % tensorboard_freq == 0:
writer.add_scalar(
'Valid_Loss_{0}'.format(tensorboard_freq),
validLossMeter.avg,
epoch * (len(valid_loader) / tensorboard_freq) +
(batch_num + 1) / tensorboard_freq)
valid_batch_bar.update(1)
valid_loss = validLossMeter.avg
writer.add_scalar('Valid_Loss_epoch', valid_loss, epoch)
logger.info("Validation Loss of epoch [{0}/{1}]: {2}\n".format(
epoch + 1, epochs, valid_loss))
# update optim scheduler
scheduler.step()
# save checkpoint
is_best = valid_loss < best_loss
best_loss_tmp = min(valid_loss, best_loss)
if not is_best:
epochs_since_improvement += 1
logger.info("Epochs since last improvement: %d\n" %
(epochs_since_improvement, ))
if epochs_since_improvement == early_stop_tolerance:
break # early stopping.
else:
epochs_since_improvement = 0
state = {
'epoch': epoch,
'loss': best_loss_tmp,
'model_state_dict': model.state_dict(),
'optimizer_state_dict': optim.state_dict(),
}
torch.save(state, ckpt_src)
logger.info("Checkpoint updated.")
best_loss = best_loss_tmp
epoch_bar.update(1)
writer.close()
batch_size=4,
shuffle=True,
num_workers=4)
classes = ('Buildings', 'MiscMan-made', 'Road', 'Track', 'Trees', 'Crops',
'Waterway', 'Standing_Water', 'Vehicle_Large', 'Vehicle_Small')
# Model definition
model = UNet(n_classes=len(classes), in_channels=_NUM_CHANNELS_)
if torch.cuda.device_count() >= 1:
print("Training model on ", torch.cuda.device_count(), "GPUs!")
model = nn.DataParallel(model)
# Loss function and Optimizer definitions
criterion = nn.BCELoss()
optimizer = optim.SGD(model.parameters(), lr=0.001, momentum=0.9)
# Network training
epoch_data = {}
for epoch in range(_NUM_EPOCHS_):
epoch_loss = 0.0
epoch_data[epoch] = {}
for i, data in enumerate(trainloader, 0):
# Get the inputs for the network
inputs = data['image'].to(_COMPUTE_DEVICE_)
labels = data['masks'].to(_COMPUTE_DEVICE_)
optimizer.zero_grad() # zero the parameter gradients
# Forward pass + Backward pass + Optimisation
outputs = model(inputs)
training_dataset = True,
in_size = 320,
train_in_size = input_size)
train_dataloader = torch.utils.data.DataLoader(SRRFDATASET, batch_size=batch_size, shuffle=True, pin_memory=True) # better than for loop
SRRFDATASET2 = ReconsDataset(all_data_path="/media/star/LuhongJin/UNC_data/SRRF/New_training_20190829/0NPY_Dataset/Dataset/Microtubule/",
maximum_intensity_4normalization_path="/home/star/0_code_lhj/DL-SIM-github/Training_codes/UNetMax_intensity.npy",
transform = ToTensor(),
training_dataset = False,
in_size = 320,
train_in_size = input_size)
validation_dataloader = torch.utils.data.DataLoader(SRRFDATASET2, batch_size=batch_size, shuffle=True, pin_memory=True) # better than for loop
model = UNet(n_channels=input_size, n_classes=output_size)
print("{} paramerters in total".format(sum(x.numel() for x in model.parameters())))
model.cuda(cuda)
optimizer = torch.optim.Adam(model.parameters(),lr=learning_rate, betas=(0.9, 0.999))
loss_all = np.zeros((2000, 4))
for epoch in range(2000):
mae_m, mae_s = val_during_training(train_dataloader)
loss_all[epoch,0] = mae_m
loss_all[epoch,1] = mae_s
mae_m, mae_s = val_during_training(validation_dataloader)
loss_all[epoch,2] = mae_m
loss_all[epoch,3] = mae_s
file = Workbook(encoding = 'utf-8')
table = file.add_sheet('loss_all')
#num_train = math.ceil(train_ratio*len(image_dataset))
#num_val = len(image_dataset) - num_train
#train_dataset, val_dataset = torch.utils.data.random_split(image_dataset, [num_train, num_val])
# Set up tensor board
writer = SummaryWriter(tfboard_dir)
# Define a loss function and optimizer
weights = torch.ones(num_class)
#Give a larger weight to SRLM for SRLM
#weights[-1] = 5
weights = weights.to(device)
#ignore_index?
#criterion = nn.CrossEntropyLoss(weights)
criterion = l.GeneralizedDiceLoss(num_classes=num_class, weight=weights)
optimizer = optim.Adam(net.parameters(), lr=c.learning_rate)
scheduler = optim.lr_scheduler.StepLR(optimizer=optimizer,
step_size=c.step_size,
gamma=0.1)
## Save and genearte patches for datasets
tp_dir = dir_names.patch_dir + "/training_data"
vp_dir = dir_names.patch_dir + "/validation_data"
#c.force_create(tp_dir)
#c.force_create(vp_dir)
#if os.path.exists(dir_names.train_patch_csv):
# os.remove(dir_names.train_patch_csv)
#if os.path.exists(dir_names.val_patch_csv):
# os.remove(dir_names.val_patch_csv)
#
train_loader = torch.utils.data.DataLoader(X_train, batch_size=batch_size, shuffle=True, pin_memory=False) # better than for loop
val_loader = torch.utils.data.DataLoader(y_train, batch_size=batch_size, shuffle=False, pin_memory=False) # better than for loop
X_train,y_train,X_test,y_test = None,None,None,None
else:
train_set = TrainDatasetFromFolder('data/DIV2K_train_HR', crop_size=CROP_SIZE, upscale_factor=UPSCALE_FACTOR)
val_set = ValDatasetFromFolder('data/DIV2K_valid_HR', crop_size=CROP_SIZE, upscale_factor=UPSCALE_FACTOR)
train_loader = DataLoader(dataset=train_set, num_workers=4, batch_size=64, shuffle=True)
val_loader = DataLoader(dataset=val_set, num_workers=4, batch_size=1, shuffle=False)
#x,y = next(iter(val_loader)),next(iter(train_loader))
#print(x[0].shape,x[1].shape,x[2].shape,y[0].shape)
#netG = Generator(UPSCALE_FACTOR,in_channels,out_channels)
netG = UNet(n_channels=in_channels, n_classes=out_channels)
#print(summary(netG,(in_channels,128,128)))
print('# generator parameters:', sum(param.numel() for param in netG.parameters()))
netD = Discriminator(out_channels)
print('# discriminator parameters:', sum(param.numel() for param in netD.parameters()))
#print(summary(netD,(out_channels,256,256)))
generator_criterion = GeneratorLoss()
#print(summary(generator_criterion,(3,256,256)))
if torch.cuda.is_available():
netG.cuda()
netD.cuda()
generator_criterion.cuda()
optimizerG = optim.Adam(netG.parameters())
optimizerD = optim.Adam(netD.parameters())
results = {'d_loss': [], 'g_loss': [], 'd_score': [], 'g_score': [], 'psnr': [], 'ssim': []}
batch_size = 1
SRRFDATASET = ReconsDataset(
test_in_path=
"/home/star/0_code_lhj/DL-SIM-github/TESTING_DATA/microtuble/HE_X2/",
transform=ToTensor(),
img_type='tif',
in_size=256)
test_dataloader = torch.utils.data.DataLoader(
SRRFDATASET, batch_size=batch_size, shuffle=True,
pin_memory=True) # better than for loop
model = UNet(n_channels=3, n_classes=1)
print("{} paramerters in total".format(
sum(x.numel() for x in model.parameters())))
model.cuda(cuda)
model.load_state_dict(
torch.load(
"/home/star/0_code_lhj/DL-SIM-github/MODELS/UNet_SIM3_microtubule.pkl"
))
model.eval()
for batch_idx, items in enumerate(test_dataloader):
image = items['image_in']
image_name = items['image_name']
print(image_name[0])
model.train()
image = np.swapaxes(image, 1, 3)
n_classes = 2
# instance_model = ReSeg(n_classes=n_classes, use_instance_seg=True, pretrained=False, usegpu=True).to(args.device)
segmenter_model = UNet(n_channels=1, n_classes=2).to(args.device)
instance_model = UNet(n_channels=1, n_classes=6).to(args.device)
loss_binary = torch.nn.BCEWithLogitsLoss()
discriminative_loss = DiscriminativeLoss(delta_var=delta_var,
delta_dist=delta_dist,
norm=2,
usegpu=True)
cross_entropy_fn = torch.nn.CrossEntropyLoss()
optimizer = torch.optim.Adam(
list(instance_model.parameters()) + list(segmenter_model.parameters()),
0.001)
def stack_iterator(n_portions, block_size, stacks=[]):
for y in range(n_portions):
for x in range(n_portions):
y1 = y * block_size
x1 = x * block_size
y2 = (y + 1) * block_size
x2 = (x + 1) * block_size
crops = [s[:, :, y1:y2, x1:x2] for s in stacks]
yield crops, (x1, y1, x2, y2)
class PG(object):
def __init__(self, configs, env):
self.configs = configs
self.env = env
self.action_size = (64, 1024)
# n_channels=3 for RGB images
# n_classes is the number of probabilities you want to get per pixel
# - For 1 class and background, use n_classes=1
# - For 2 classes, use n_classes=1
# - For N > 2 classes, use n_classes=N
# TODO now I assume input<->output size are equal, which might not be true, so we need some modifications onto Unet if necesary
self.actor = UNet(
n_channels=3, n_classes=1, bilinear=True
) # [B,C, H_in=372, W_in=1242] -> [B, C, H_out=64, W_out=1024]
self.optimizer = Adam(self.actor.parameters(), lr=configs['lr'])
self.actor.to(device)
def get_action(self, state, deterministic=False):
"""Given the state, produces an action, the probability of the action, the log probability of the action, and
the argmax action"""
action_probabilities = self.actor(
state) # output size should be [B*H*W]
action_probabilities = torch.sigmoid(
action_probabilities) # make sure the probs are in range [0,1]
# B, _, _, _ = action_probabilities.shape
action_probabilities = action_probabilities[:, :, :self.
action_size[0], :self.
action_size[1]]
action_probabilities = torch.squeeze(action_probabilities, 1)
# assert action_probabilities.size()[1, 2] == self.action_size, "Actor output the wrong size"
# action_probabilities_flat = action_probabilities.contiguous().view(B, -1)
# TODO leave this to future process; seems it will get the index
max_probability_action = torch.argmax(action_probabilities, dim=-1)
if deterministic:
# using deteministic policy during test time
action = action_probabilities(action_probabilities > 0.5).cpu()
else:
# using stochastic policy during traning time
action_distribution = Bernoulli(
action_probabilities
) # this creates a distribution to sample from
action = action_distribution.sample().cpu(
) # sample the discrete action and copy it to cpu
# Have to deal with situation of 0.0 probabilities because we can't do log 0
z = action_probabilities == 0.0
z = z.float() * 1e-8
log_action_probabilities = torch.log(action_probabilities + z)
return action, action_probabilities, log_action_probabilities, max_probability_action
def compute_loss(self, obs, act, rew):
"""make loss function whose gradient, for the right data, is policy gradient"""
# TODO we may do not need to calculate it for the second time.
act_baseline, _, logp, _ = self.get_action(obs, deterministic=True)
# advantage
_, rew_baseline, _, _ = self.env.step(act_baseline, obs=obs)
advantage = rew.to(device).float() - rew_baseline.to(device).float()
loss = logp * Variable(advantage).expand_as(act)
loss = loss.mean()
return loss
def update(self, batch_obs, batch_acts, batch_rews):
"""take a single policy gradient update step for a batch"""
self.optimizer.zero_grad()
batch_loss = self.compute_loss(
obs=torch.as_tensor(batch_obs, dtype=torch.float32),
act=torch.as_tensor(batch_acts, dtype=torch.int32),
rew=torch.as_tensor(batch_rews, dtype=torch.int32),
)
batch_loss.backward()
self.optimizer.step()
return batch_loss
if convert_to_2d:
inp_set = get_2d_converted_data(inp_set)
inp_set = torch.from_numpy(inp_set).float()
file_name = os.path.basename(inp_file)
out_file = os.path.join(out_dir, file_name)
data.append((inp_set, out_file))
return data
def save_pred(model, data):
model.eval()
for image, file_path in data:
img = image.cuda(cuda)
pred = model(img)
pred = pred.detach().cpu().numpy()[0]
pred = (pred * 255) .astype(np.uint8)
# save_path = file_path.replace('.mat', '.png')
# cv2.imwrite(save_path, pred)
pred = pred.transpose((1, 2, 0))
savemat(file_path, {'crop_g': pred})
if __name__ == '__main__':
cuda = torch.device('cuda')
model = UNet(n_channels=45, n_classes=3)
print("{} Parameters in total".format(sum(x.numel() for x in model.parameters())))
model.cuda(cuda)
model.load_state_dict(torch.load(model_loc+"Model_Final_999_3_5.pkl"))
model.eval()
model.cuda(cuda)
data = get_images()
save_pred(model, data)
class UNetObjPrior(nn.Module):
"""
Wrapper around UNet that takes object priors (gaussians) and images
as input.
"""
def __init__(self, params, depth=5):
super(UNetObjPrior, self).__init__()
self.in_channels = 4
self.model = UNet(1, self.in_channels, depth, cuda=params['cuda'])
self.params = params
self.device = torch.device('cuda' if params['cuda'] else 'cpu')
def forward(self, im, obj_prior):
x = torch.cat((im, obj_prior), dim=1)
return self.model(x)
def train(self, dataloader_train, dataloader_val):
since = time.time()
best_loss = float("inf")
dataloader_train.mode = 'train'
dataloader_val.mode = 'val'
dataloaders = {'train': dataloader_train, 'val': dataloader_val}
optimizer = optim.SGD(self.model.parameters(),
momentum=self.params['momentum'],
lr=self.params['lr'],
weight_decay=self.params['weight_decay'])
train_logger = LossLogger('train', self.params['batch_size'],
len(dataloader_train),
self.params['out_dir'])
val_logger = LossLogger('val', self.params['batch_size'],
len(dataloader_val), self.params['out_dir'])
loggers = {'train': train_logger, 'val': val_logger}
# self.criterion = WeightedMSE(dataloader_train.get_classes_weights(),
# cuda=self.params['cuda'])
self.criterion = nn.MSELoss()
for epoch in range(self.params['num_epochs']):
print('Epoch {}/{}'.format(epoch, self.params['num_epochs'] - 1))
print('-' * 10)
# Each epoch has a training and validation phase
for phase in ['train', 'val']:
if phase == 'train':
#scheduler.step()
self.model.train()
else:
self.model.eval() # Set model to evaluate mode
running_loss = 0.0
running_corrects = 0
# Iterate over data.
samp = 1
for i, data in enumerate(dataloaders[phase]):
# zero the parameter gradients
optimizer.zero_grad()
# forward
# track history if only in train
with torch.set_grad_enabled(phase == 'train'):
out = self.forward(data.image, data.obj_prior)
loss = self.criterion(out, data.truth)
# backward + optimize only if in training phase
if phase == 'train':
loss.backward()
optimizer.step()
loggers[phase].update(epoch, samp, loss.item())
samp += 1
loggers[phase].print_epoch(epoch)
# Generate train prediction for check
if phase == 'train':
path = os.path.join(self.params['out_dir'], 'previews',
'epoch_{:04d}.jpg'.format(epoch))
data = dataloaders['val'].sample_uniform()
pred = self.forward(data.image, data.obj_prior)
im_ = data.image[0]
truth_ = data.truth[0]
pred_ = pred[0, ...]
utls.save_tensors(im_, pred_, truth_, path)
if phase == 'val' and (loggers['val'].get_loss(epoch) <
best_loss):
best_loss = loggers['val'].get_loss(epoch)
loggers[phase].save('log_{}.csv'.format(phase))
# save checkpoint
if phase == 'val':
is_best = loggers['val'].get_loss(epoch) <= best_loss
path = os.path.join(self.params['out_dir'],
'checkpoint.pth.tar')
utls.save_checkpoint(
{
'epoch': epoch + 1,
'state_dict': self.model.state_dict(),
'best_loss': best_loss,
'optimizer': optimizer.state_dict()
},
is_best,
path=path)
def load_checkpoint(self, path, device='gpu'):
if (device != 'gpu'):
checkpoint = torch.load(path,
map_location=lambda storage, loc: storage)
else:
checkpoint = torch.load(path)
self.model.load_state_dict(checkpoint['state_dict'])
LE_512 = cropImage(LE_img, IMG_SHAPE[0], IMG_SHAPE[1])
sample_le = {}
for le_512 in LE_512:
tiles = crop_prepare(le_512, CROP_STEP, IMG_SIZE)
for n, img in enumerate(tiles):
if n not in sample_le:
sample_le[n] = []
img = transform.resize(img, (IMG_SIZE * 2, IMG_SIZE * 2),
preserve_range=True,
order=3)
sample_le[n].append(img)
SC_UNET = UNet(n_channels=15, n_classes=1)
print("{} paramerters in total".format(
sum(x.numel() for x in SC_UNET.parameters())))
SC_UNET.cuda(cuda)
SC_UNET.load_state_dict(torch.load(model_path))
# SC_UNET.load_state_dict(torch.load(os.path.join(dir_path,"model","HE_HER_mito","HE_X2_HER_0825.pkl")))
SC_UNET.eval()
SRRFDATASET = ReconsDataset(img_dict=sample_he,
transform=ToTensor(),
in_norm=LE_in_norm,
img_type=".tif",
in_size=256)
test_dataloader = torch.utils.data.DataLoader(
SRRFDATASET, batch_size=1, shuffle=False,
pin_memory=True) # better than for loop
result = np.zeros((256, 256, len(SRRFDATASET)))
for batch_idx, items in enumerate(test_dataloader):
class Train(object):
def __init__(self, configs):
self.batch_size = configs.get("batch_size", "16")
self.epochs = configs.get("epochs", "100")
self.lr = configs.get("lr", "0.0001")
device_args = configs.get("device", "cuda")
self.device = torch.device(
"cpu" if not torch.cuda.is_available() else device_args)
self.workers = configs.get("workers", "4")
self.vis_images = configs.get("vis_images", "200")
self.vis_freq = configs.get("vis_freq", "10")
self.weights = configs.get("weights", "./weights")
if not os.path.exists(self.weights):
os.mkdir(self.weights)
self.logs = configs.get("logs", "./logs")
if not os.path.exists(self.weights):
os.mkdir(self.weights)
self.images_path = configs.get("images_path", "./data")
self.is_resize = config.get("is_resize", False)
self.image_short_side = config.get("image_short_side", 256)
self.is_padding = config.get("is_padding", False)
is_multi_gpu = config.get("DateParallel", False)
pre_train = config.get("pre_train", False)
model_path = config.get("model_path", './weights/unet_idcard_adam.pth')
# self.image_size = configs.get("image_size", "256")
# self.aug_scale = configs.get("aug_scale", "0.05")
# self.aug_angle = configs.get("aug_angle", "15")
self.step = 0
self.dsc_loss = DiceLoss()
self.model = UNet(in_channels=Dataset.in_channels,
out_channels=Dataset.out_channels)
if pre_train:
self.model.load_state_dict(torch.load(model_path,
map_location=self.device),
strict=False)
if is_multi_gpu:
self.model = nn.DataParallel(self.model)
self.model.to(self.device)
self.best_validation_dsc = 0.0
self.loader_train, self.loader_valid = self.data_loaders()
self.params = [p for p in self.model.parameters() if p.requires_grad]
self.optimizer = optim.Adam(self.params,
lr=self.lr,
weight_decay=0.0005)
# self.optimizer = torch.optim.SGD(self.params, lr=self.lr, momentum=0.9, weight_decay=0.0005)
self.scheduler = lr_scheduler.LR_Scheduler_Head(
'poly', self.lr, self.epochs, len(self.loader_train))
def datasets(self):
train_datasets = Dataset(
images_dir=self.images_path,
# image_size=self.image_size,
subset="train", # train
transform=get_transforms(train=True),
is_resize=self.is_resize,
image_short_side=self.image_short_side,
is_padding=self.is_padding)
# valid_datasets = train_datasets
valid_datasets = Dataset(
images_dir=self.images_path,
# image_size=self.image_size,
subset="validation", # validation
transform=get_transforms(train=False),
is_resize=self.is_resize,
image_short_side=self.image_short_side,
is_padding=False)
return train_datasets, valid_datasets
def data_loaders(self):
dataset_train, dataset_valid = self.datasets()
loader_train = DataLoader(
dataset_train,
batch_size=self.batch_size,
shuffle=True,
drop_last=True,
num_workers=self.workers,
)
loader_valid = DataLoader(
dataset_valid,
batch_size=1,
drop_last=False,
num_workers=self.workers,
)
return loader_train, loader_valid
@staticmethod
def dsc_per_volume(validation_pred, validation_true):
assert len(validation_pred) == len(validation_true)
dsc_list = []
for p in range(len(validation_pred)):
y_pred = np.array([validation_pred[p]])
y_true = np.array([validation_true[p]])
dsc_list.append(dsc(y_pred, y_true))
return dsc_list
@staticmethod
def get_logger(filename, verbosity=1, name=None):
level_dict = {0: logging.DEBUG, 1: logging.INFO, 2: logging.WARNING}
formatter = logging.Formatter(
"[%(asctime)s][%(filename)s][line:%(lineno)d][%(levelname)s] %(message)s"
)
logger = logging.getLogger(name)
logger.setLevel(level_dict[verbosity])
fh = logging.FileHandler(filename, "w")
fh.setFormatter(formatter)
logger.addHandler(fh)
sh = logging.StreamHandler()
sh.setFormatter(formatter)
logger.addHandler(sh)
return logger
def train_one_epoch(self, epoch):
self.model.train()
loss_train = []
for i, data in enumerate(self.loader_train):
self.scheduler(self.optimizer, i, epoch, self.best_validation_dsc)
x, y_true = data
x, y_true = x.to(self.device), y_true.to(self.device)
y_pred = self.model(x)
# print('1111', y_pred.size())
# print('2222', y_true.size())
loss = self.dsc_loss(y_pred, y_true)
loss_train.append(loss.item())
self.optimizer.zero_grad()
loss.backward()
self.optimizer.step()
# lr_scheduler.step()
if self.step % 200 == 0:
print('Epoch:[{}/{}]\t iter:[{}]\t loss={:.5f}\t '.format(
epoch, self.epochs, i, loss))
self.step += 1
def eval_model(self, patience):
self.model.eval()
loss_valid = []
validation_pred = []
validation_true = []
# early_stopping = EarlyStopping(patience=patience, verbose=True)
for i, data in enumerate(self.loader_valid):
x, y_true = data
x, y_true = x.to(self.device), y_true.to(self.device)
# print(x.size())
# print(333,x[0][2])
with torch.no_grad():
y_pred = self.model(x)
loss = self.dsc_loss(y_pred, y_true)
# print(y_pred.shape)
mask = y_pred > 0.5
mask = mask * 255
mask = mask.cpu().numpy()[0][0]
# print(mask)
# print(mask.shape())
cv2.imwrite('result.png', mask)
loss_valid.append(loss.item())
y_pred_np = y_pred.detach().cpu().numpy()
validation_pred.extend(
[y_pred_np[s] for s in range(y_pred_np.shape[0])])
y_true_np = y_true.detach().cpu().numpy()
validation_true.extend(
[y_true_np[s] for s in range(y_true_np.shape[0])])
# early_stopping(loss_valid, self.model)
# if early_stopping.early_stop:
# print('Early stopping')
# import sys
# sys.exit(1)
mean_dsc = np.mean(
self.dsc_per_volume(
validation_pred,
validation_true,
))
# print('mean_dsc:', mean_dsc)
if mean_dsc > self.best_validation_dsc:
self.best_validation_dsc = mean_dsc
torch.save(self.model.state_dict(),
os.path.join(self.weights, "unet_xia_adam.pth"))
print("Best validation mean DSC: {:4f}".format(
self.best_validation_dsc))
def main(self):
# print('train is begin.....')
# print('load data end.....')
# loaders = {"train": loader_train, "valid": loader_valid}
for epoch in tqdm(range(self.epochs), total=self.epochs):
self.train_one_epoch(epoch)
self.eval_model(patience=10)
torch.save(self.model.state_dict(),
os.path.join(self.weights, "unet_final.pth"))
SRRFDATASET2 = ReconsDataset(
train_in_path=
"/media/star/LuhongJin/UNC_data/SIM/ALL_data/microtubule/Training_Testing/testing_HE_X2/",
train_gt_path=
"/media/star/LuhongJin/UNC_data/SIM/ALL_data/microtubule/Training_Testing/testing_HER/",
transform=ToTensor(),
img_type='tif',
in_size=256)
validation_dataloader = torch.utils.data.DataLoader(
SRRFDATASET2, batch_size=batch_size, shuffle=True,
pin_memory=True) # better than for loop
model = UNet(n_channels=15, n_classes=1)
print("{} paramerters in total".format(
sum(x.numel() for x in model.parameters())))
model.cuda(cuda)
optimizer = torch.optim.Adam(model.parameters(),
lr=learning_rate,
betas=(0.9, 0.999))
loss_all = np.zeros((2000, 4))
for epoch in range(2000):
mae_m, mae_s = val_during_training(train_dataloader)
loss_all[epoch, 0] = mae_m
loss_all[epoch, 1] = mae_s
mae_m, mae_s = val_during_training(validation_dataloader)
loss_all[epoch, 2] = mae_m
loss_all[epoch, 3] = mae_s
#train_loader = torch.utils.data.DataLoader(train_data, batch_size=batch_size, shuffle=True, pin_memory=False) # better than for loop
#val_loader = torch.utils.data.DataLoader(test_data, batch_size=batch_size, shuffle=False, pin_memory=False) # better than for loop
#torch.autograd.set_detect_anomaly(True)
if True: # __name__ == '__main__':
#opt = parser.parse_args()
#train_set = TrainDatasetFromFolder('data/DIV2K_train_HR', crop_size=CROP_SIZE, upscale_factor=UPSCALE_FACTOR)
#val_set = ValDatasetFromFolder('data/DIV2K_valid_HR', upscale_factor=UPSCALE_FACTOR)
#train_loader = DataLoader(dataset=train_set, num_workers=4, batch_size=64, shuffle=True)
#val_loader = DataLoader(dataset=val_set, num_workers=4, batch_size=1, shuffle=False)
netG = UNet(n_channels=15, n_classes=1)
#print(summary(netG,(15,128,128)))
print('# generator parameters:',
sum(param.numel() for param in netG.parameters()))
netD = Discriminator()
print('# discriminator parameters:',
sum(param.numel() for param in netD.parameters()))
#print(summary(netD,(1,256,256)))
generator_criterion = GeneratorLoss()
if torch.cuda.is_available():
netG.cuda()
netD.cuda()
generator_criterion.cuda()
optimizerG = optim.Adam(netG.parameters())
optimizerD = optim.Adam(netD.parameters())
nn.init.constant_(m.bias.data, 0)
net = UNet(3, 1).to(device)
# Apply the weights_init function to randomly initialize all weights
net.apply(weights_init)
print(net)
print('parameters:', get_layer_param(net))
# Initialize BCELoss function
criterion = nn.MSELoss()
# Setup Adam optimizers for both G and D
# optimizer = optim.Adam(net.parameters(), lr=lr)
optimizer = optim.Adam(net.parameters(), lr=lr, betas=(0.5, 0.999))
# optimizer = optim.SGD(net.parameters(), lr=lr, momentum=0.9, weight_decay=0.0005)
# define a method to save loss image
def save_loss_image(list_loss, path):
plt.figure(figsize=(10, 5))
plt.title("Generator and Discriminator Loss During Training")
plt.plot(list_loss, label="loss")
plt.xlabel("iterations")
plt.ylabel("Loss")
plt.legend()
plt.savefig(os.path.join(path, 'loss.jpg'))
plt.close()
list_loss = []
# Load model if available
if(resume==True):
print('Resuming training....')
generator.load_state_dict(torch.load(os.path.join(model_path,'model_gen_latest')))
discriminator_g.load_state_dict(torch.load(os.path.join(model_path,'model_gdis_latest')))
discriminator_l.load_state_dict(torch.load(os.path.join(model_path,'model_ldis_latest')))
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
generator = generator.to(device)
discriminator_g = discriminator_g.to(device)
discriminator_l = discriminator_l.to(device)
optimizer_g = optim.Adam(discriminator_g.parameters(), lr=0.00005)
optimizer_l = optim.Adam(discriminator_l.parameters(), lr=0.00005)
gen_optimizer = optim.Adam(generator.parameters(), lr=0.0002)
lossdis = nn.BCELoss()
lossgen = FocalLoss()
lamda = 75
data_loader = load_images(data_path)
num_epochs = 2000
for epoch in range(num_epochs):
print()
for n_batch, (real_data, gt_data) in enumerate(data_loader):
# 1. Train Discriminator
N = real_data.size(0)
return model
if __name__ == '__main__':
lr = 0.001
model = UNet(n_channels=1)
#num_ftrs = model.fc.in_features
# Here the size of each output sample is set to 2
# Alternatively it can be generalized to nn.Linear(num_ftrs, len(class_names))
#model.fc = nn.Linear(num_ftrs, 2)
model = model.to(device)
criterion = nn.CrossEntropyLoss()
# Observe that all parameters are being optimized
optimizer_ft = optim.Adam(model.parameters(), lr=lr) #, momentum=0.9)
# Decay LR by a factor of 0.1 every 7 epochs
exp_lr_scheduler = lr_scheduler.StepLR(optimizer_ft,
step_size=7,
gamma=0.1)
# Finetune training the convnet and evaluation
model = train_model(model,
criterion,
optimizer_ft,
exp_lr_scheduler,
num_epochs=25)
writer = SummaryWriter()
image = Image.open('./ht2-c2.jpg')
out = TF.to_tensor(image)
out = out.reshape(1, 3, 640, 640)
inp = torch.rand(1, 3, 640, 640)
fig = plt.figure()
plt.imshow(out[0].permute(1, 2, 0).numpy())
# plt.show
writer.add_figure("Ground Truth", fig)
fig = plt.figure()
plt.imshow(inp[0].permute(1, 2, 0).numpy())
writer.add_figure("Input", fig)
num_iter = 500
writer.add_scalar("Number_of_Iterations", num_iter)
model = UNet(3, 3)
if torch.cuda.is_available():
model.cuda()
criterion = nn.MSELoss()
learning_rate = 0.1
writer.add_scalar("Learning_Rate", learning_rate)
optimizer = torch.optim.SGD(model.parameters(), lr=learning_rate)
train(num_iter, inp, out, model, optimizer, criterion)
writer.close()
num_workers=1,
pin_memory=True)
# logging training overview
print('-----\n Start training:')
print(
f'epochs: {args.epochs} \t batch size: {args.batch_size} \t learning rate: {args.learning_rate} \t'
)
print(
f'training size: {n_train} \t validation size: {n_val} \t checkpoints_dir: {args.checkpoints_dir} \t images downscale: {args.down_scale}'
)
print('-----')
## --- Set up training
global_step = 0
optimizer = optim.RMSprop(net.parameters(),
lr=args.learning_rate,
weight_decay=1e-8)
if net.n_classes > 1:
criterion = nn.CrossEntropyLoss()
else:
criterion = nn.BCEWithLogitsLoss()
## --- Start training
epoch_loss_list = []
val_score_list = []
num_batches_per_epoch = len(dataset) // args.batch_size
for epoch in range(args.epochs):
net.train()
epoch_loss = 0
