Skip to content

Minor fixes #4

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 1 commit into
base: master
Choose a base branch
from
Open

Minor fixes #4

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 2 additions & 2 deletions README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -32,7 +32,7 @@ Alternatively you can build your own dataset by setting up the following directo
| | └── test # Testing
| | | ├── A # Contains domain A images (i.e. Bruce Wayne)
| | | └── B # Contains domain B images (i.e. Batman)

### 2. Train!
```
./train --dataroot datasets/<dataset_name>/ --cuda
Expand All @@ -43,7 +43,7 @@ Both generators and discriminators weights will be saved under the output direct

If you don't own a GPU remove the --cuda option, although I advise you to get one!

You can also view the training progress as well as live output images by running ```python3 -m visdom``` in another terminal and opening [http://localhost:8097/](http://localhost:8097/) in your favourite web browser. This should generate training loss progress as shown below (default params, horse2zebra dataset):
You can also view the training progress as well as live output images by running ```python3 -m visdom.server``` in another terminal and opening [http://localhost:8097/](http://localhost:8097/) in your favourite web browser. This should generate training loss progress as shown below (default params, horse2zebra dataset):

![Generator loss](https://github.com/ai-tor/PyTorch-CycleGAN/raw/master/output/loss_G.png)
![Discriminator loss](https://github.com/ai-tor/PyTorch-CycleGAN/raw/master/output/loss_D.png)
Expand Down
16 changes: 8 additions & 8 deletions train
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -72,19 +72,19 @@ lr_scheduler_D_B = torch.optim.lr_scheduler.LambdaLR(optimizer_D_B, lr_lambda=La
Tensor = torch.cuda.FloatTensor if opt.cuda else torch.Tensor
input_A = Tensor(opt.batchSize, opt.input_nc, opt.size, opt.size)
input_B = Tensor(opt.batchSize, opt.output_nc, opt.size, opt.size)
target_real = Variable(Tensor(opt.batchSize).fill_(1.0), requires_grad=False)
target_fake = Variable(Tensor(opt.batchSize).fill_(0.0), requires_grad=False)
target_real = Variable(Tensor(opt.batchSize,1).fill_(1.0), requires_grad=False)
target_fake = Variable(Tensor(opt.batchSize,1).fill_(0.0), requires_grad=False)

fake_A_buffer = ReplayBuffer()
fake_B_buffer = ReplayBuffer()

# Dataset loader
transforms_ = [ transforms.Resize(int(opt.size*1.12), Image.BICUBIC),
transforms.RandomCrop(opt.size),
transforms_ = [ transforms.Resize(int(opt.size*1.12), Image.BICUBIC),
transforms.RandomCrop(opt.size),
transforms.RandomHorizontalFlip(),
transforms.ToTensor(),
transforms.Normalize((0.5,0.5,0.5), (0.5,0.5,0.5)) ]
dataloader = DataLoader(ImageDataset(opt.dataroot, transforms_=transforms_, unaligned=True),
dataloader = DataLoader(ImageDataset(opt.dataroot, transforms_=transforms_, unaligned=True),
batch_size=opt.batchSize, shuffle=True, num_workers=opt.n_cpu)

# Loss plot
Expand Down Expand Up @@ -128,7 +128,7 @@ for epoch in range(opt.epoch, opt.n_epochs):
# Total loss
loss_G = loss_identity_A + loss_identity_B + loss_GAN_A2B + loss_GAN_B2A + loss_cycle_ABA + loss_cycle_BAB
loss_G.backward()

optimizer_G.step()
###################################

Expand Down Expand Up @@ -157,7 +157,7 @@ for epoch in range(opt.epoch, opt.n_epochs):
# Real loss
pred_real = netD_B(real_B)
loss_D_real = criterion_GAN(pred_real, target_real)

# Fake loss
fake_B = fake_B_buffer.push_and_pop(fake_B)
pred_fake = netD_B(fake_B.detach())
Expand All @@ -172,7 +172,7 @@ for epoch in range(opt.epoch, opt.n_epochs):

# Progress report (http://localhost:8097)
logger.log({'loss_G': loss_G, 'loss_G_identity': (loss_identity_A + loss_identity_B), 'loss_G_GAN': (loss_GAN_A2B + loss_GAN_B2A),
'loss_G_cycle': (loss_cycle_ABA + loss_cycle_BAB), 'loss_D': (loss_D_A + loss_D_B)},
'loss_G_cycle': (loss_cycle_ABA + loss_cycle_BAB), 'loss_D': (loss_D_A + loss_D_B)},
images={'real_A': real_A, 'real_B': real_B, 'fake_A': fake_A, 'fake_B': fake_B})

# Update learning rates
Expand Down