def valid(valid_loader, model, epoch, logger):
model.eval() # eval mode (dropout and batchnorm is NOT used)
losses = AverageMeter()
# Batches
for i, (img, alpha_label) in enumerate(valid_loader):
# Move to GPU, if available
img = img.type(torch.FloatTensor).to(device) # [N, 3, 320, 320]
alpha_label = alpha_label.type(torch.FloatTensor).to(device) # [N, 320, 320]
alpha_label = alpha_label.reshape((-1, 2, im_size * im_size)) # [N, 320*320]
# Forward prop.
alpha_out = model(img) # [N, 320, 320]
alpha_out = alpha_out.reshape((-1, 1, im_size * im_size)) # [N, 320*320]
# Calculate loss
# loss = criterion(alpha_out, alpha_label)
loss = alpha_prediction_loss(alpha_out, alpha_label)
# Keep track of metrics
losses.update(loss.item())
if i % print_freq == 0:
status = 'Epoch: [{0}][{1}/{2}]\t' \
'Loss {loss.val:.4f} ({loss.avg:.4f})\t'.format(epoch, i, len(valid_loader), loss=losses)
logger.info(status)
# Print status
status = 'Validation: Loss {loss.avg:.4f}\n'.format(loss=losses)
logger.info(status)
return losses.avg
def valid(valid_loader, model, logger):
model.eval() # eval mode (dropout and batchnorm is NOT used)
losses = AverageMeter()
# Batches
for img, alpha_label in tqdm(valid_loader):
# Move to GPU, if available
img = img.type(torch.FloatTensor).to(device) # [N, 4, 320, 320]
alpha_label = alpha_label.type(torch.FloatTensor).to(device) # [N, 2, 320, 320]
alpha_label = alpha_label.reshape((-1, 2, im_size * im_size)) # [N, 2, 320*320]
# Forward prop.
alpha_out = model(img) # [N, 320, 320]
alpha_out = alpha_out.reshape((-1, 1, im_size * im_size)) # [N, 320*320]
# Calculate loss
# loss = criterion(alpha_out, alpha_label)
loss = alpha_prediction_loss(alpha_out, alpha_label)
# Keep track of metrics
losses.update(loss.item())
# Print status
status = 'Validation: Loss {loss.avg:.4f}\n'.format(loss=losses)
logger.info(status)
return losses.avg
def valid(valid_loader, model, logger, device):
model.eval() # eval mode (dropout and batchnorm is NOT used)
losses = AverageMeter()
with torch.no_grad():
para_valid_loader = pl.ParallelLoader(valid_loader, [device]).per_device_loader(device)
for i, batch in enumerate(para_valid_loader):
img, alpha_label = batch
# # Batches
# for img, alpha_label in valid_loader:
# Move to GPU, if available
img = img.type(torch.FloatTensor).to(device) # [N, 3, 320, 320]
alpha_label = alpha_label.type(torch.FloatTensor).to(device) # [N, 320, 320]
alpha_label = alpha_label.reshape((-1, 2, im_size * im_size)) # [N, 320*320]
# Forward prop.
alpha_out = model(img) # [N, 320, 320]
alpha_out = alpha_out.reshape((-1, 1, im_size * im_size)) # [N, 320*320]
# Calculate loss
# loss = criterion(alpha_out, alpha_label)
loss = alpha_prediction_loss(alpha_out, alpha_label)
# Keep track of metrics
losses.update(loss.item())
# Print status
status = 'Validation: Loss {loss.avg:.4f}\n'.format(loss=losses)
logger.info(status)
return losses.avg
def train(train_loader, model, optimizer, epoch, logger):
torch.nn.Module.dump_patches = True
model.train() # train mode (dropout and batchnorm is used)
losses = AverageMeter()
# Batches
for i, (img, alpha_label) in enumerate(train_loader):
# Move to GPU, if available
img = img.type(torch.FloatTensor).to(device) # [N, 4, 320, 320]
alpha_label = alpha_label.type(torch.FloatTensor).to(
device) # [N, 320, 320]
alpha_label = alpha_label.reshape(
(-1, 2, im_size * im_size)) # [N, 320*320]
# Forward prop.
alpha_out = model(img) # [N, 3, 320, 320]
alpha_out = alpha_out.reshape(
(-1, 1, im_size * im_size)) # [N, 320*320]
# Calculate loss
# loss = criterion(alpha_out, alpha_label)
loss = alpha_prediction_loss(alpha_out, alpha_label)
# Back prop.
optimizer.zero_grad()
loss.backward()
# Clip gradients
clip_gradient(optimizer, grad_clip)
# Update weights
optimizer.step()
# Keep track of metrics
losses.update(loss.item())
# Print status
if i % print_freq == 0:
status = 'Epoch: [{0}][{1}/{2}]\t' \
'Loss {loss.val:.4f} ({loss.avg:.4f})\t'.format(epoch, i, len(train_loader), loss=losses)
logger.info(status)
return losses.avg
def train(train_loader, model, optimizer, epoch, logger):
model.train() # train mode (dropout and batchnorm is used)
losses = AverageMeter()
# Batches
for i, (img, alpha_label, image, fg, bg) in enumerate(train_loader):
# Move to GPU, if available
img = img.type(torch.FloatTensor).to(device) # [N, 4, 320, 320]
alpha_label = alpha_label.type(torch.FloatTensor).to(device) # [N, 2, 320, 320]
image = image.type(torch.FloatTensor).to(device)
fg = fg.type(torch.FloatTensor).to(device)
bg = bg.type(torch.FloatTensor).to(device)
# Forward prop.
alpha_out = model(img) # [N, 320, 320]
# Calculate loss
# loss = criterion(alpha_out, alpha_label)
alpha_loss = alpha_prediction_loss(alpha_out, alpha_label)
comp_loss = composition_loss(alpha_out, alpha_label, image, fg, bg)
w_l = 0.5
loss = w_l * alpha_loss + (1 - w_l) * comp_loss
# Back prop.
optimizer.zero_grad()
loss.backward()
# Clip gradients
clip_gradient(optimizer, grad_clip)
# Update weights
optimizer.step()
# Keep track of metrics
losses.update(loss.item())
# Print status
if i % print_freq == 0:
status = 'Epoch: [{0}][{1}/{2}]\t' \
'Loss {loss.val:.4f} ({loss.avg:.4f})\t'.format(epoch, i, len(train_loader), loss=losses)
logger.info(status)
return losses.avg
def valid(valid_loader, model, epoch, logger):
model.eval() # eval mode (dropout and batchnorm is NOT used)
losses = AverageMeter()
# Batches
for i, (img, alpha_label, image, fg, bg) in enumerate(valid_loader):
# Move to GPU, if available
img = img.type(torch.FloatTensor).to(device) # [N, 4, 320, 320]
alpha_label = alpha_label.type(torch.FloatTensor).to(device) # [N, 2, 320, 320]
image = image.type(torch.FloatTensor).to(device)
fg = fg.type(torch.FloatTensor).to(device)
bg = bg.type(torch.FloatTensor).to(device)
# Forward prop.
alpha_out = model(img) # [N, 320, 320]
# Calculate loss
# loss = criterion(alpha_out, alpha_label)
alpha_loss = alpha_prediction_loss(alpha_out, alpha_label)
comp_loss = composition_loss(alpha_out, alpha_label, image, fg, bg)
w_l = 0.5
loss = w_l * alpha_loss + (1 - w_l) * comp_loss
# Keep track of metrics
losses.update(loss.item())
if i % print_freq == 0:
status = 'Epoch: [{0}][{1}/{2}]\t' \
'Loss {loss.val:.4f} ({loss.avg:.4f})\t'.format(epoch, i, len(valid_loader), loss=losses)
logger.info(status)
# Print status
status = 'Validation: Loss {loss.avg:.4f}\n'.format(loss=losses)
logger.info(status)
return losses.avg
