rm wrong file

edit net a bit
This commit is contained in:
lukas-heiligenbrunner 2022-07-11 23:36:07 +02:00
parent 0f0c789981
commit 3b4ef675ad
4 changed files with 44 additions and 19 deletions

View File

@ -80,17 +80,17 @@ def get_image_loader(path: str, precision: np.float32 or np.float64):
transform_chain=transforms.Compose([transforms.RandomHorizontalFlip(),
transforms.RandomVerticalFlip()]),
precision=precision)
image_dataset_augmented = ImageDataset(path,
offsetrange=(0, 8),
spacingrange=(2, 6),
transform_chain=transforms.Compose([transforms.RandomHorizontalFlip(),
transforms.RandomVerticalFlip(),
transforms.GaussianBlur(3, 4)]),
precision=precision)
#
# image_dataset_augmented = ImageDataset(path,
# offsetrange=(0, 8),
# spacingrange=(2, 6),
# transform_chain=transforms.Compose([transforms.RandomHorizontalFlip(),
# transforms.RandomVerticalFlip(),
# transforms.GaussianBlur(3, 4)]),
# precision=precision)
# merge different datasets here!
merged_dataset = torch.utils.data.ConcatDataset([image_dataset, image_dataset_augmented])
merged_dataset = torch.utils.data.ConcatDataset([image_dataset])
totlen = len(merged_dataset)
test_set_size = .1

View File

@ -48,13 +48,22 @@ def train_model():
print(f"Epoch {epoch}/{n_epochs}\n")
i = 0
for input_tensor, mask, target_tensor in train_loader:
input_tensor = input_tensor.to(device)
mask = mask.to(device)
target_tensor = target_tensor.to(device)
optimizer.zero_grad() # reset gradients
input_tensor = torch.cat((input_tensor, mask), 1)
output = nn(input_tensor.to(device)) # get model output (forward pass)
output = nn(input_tensor) # get model output (forward pass)
loss = loss_function(output.to(device),
target_tensor.to(device)) # compute loss given model output and true target
output_flat = output * (1 - mask)
output_flat = output_flat[1 - mask > 0]
rest = target_tensor * (1 - mask)
rest = rest[1 - mask > 0]
loss = loss_function(output_flat, rest) # compute loss given model output and true target
loss.backward() # compute gradients (backward pass)
optimizer.step() # perform gradient descent update step
@ -101,7 +110,14 @@ def eval_model(model: torch.nn.Module, dataloader: torch.utils.data.DataLoader,
input = torch.cat((input, mask), 1)
out = model(input)
loss += loss_fn(out, target).item()
out = out * (1 - mask)
out = out[1 - mask > 0]
rest = target * (1 - mask)
rest = rest[1 - mask > 0]
loss += loss_fn(out, rest).item()
print(f'\rEval prog[{i}/{len(dataloader) * dataloader.batch_size}]', end='')
i += dataloader.batch_size
print()
@ -112,9 +128,9 @@ def eval_model(model: torch.nn.Module, dataloader: torch.utils.data.DataLoader,
def plot(inputs, targets, predictions, path, update, epoch):
"""Plotting the inputs, targets and predictions to file `path`"""
os.makedirs(path, exist_ok=True)
fig, axes = plt.subplots(ncols=3, figsize=(15, 5))
fig, axes = plt.subplots(ncols=4, figsize=(15, 5))
for ax, data, title in zip(axes, [inputs, targets, predictions], ["Input", "Target", "Prediction"]):
for ax, data, title in zip(axes, [inputs, targets, predictions, predictions-targets], ["Input", "Target", "Prediction", "diff"]):
ax.clear()
ax.set_title(title)
ax.imshow(DataLoader.postprocess(np.transpose(data[:3, :, :], (1, 2, 0))), interpolation="none")

12
Net.py
View File

@ -7,18 +7,24 @@ from torch import nn
class ImageNN(torch.nn.Module):
def __init__(self, precision: np.float32 or np.float64, n_in_channels: int = 1, n_hidden_layers: int = 3,
n_kernels: int = 32, kernel_size: int = 7):
n_kernels: int = 32, kernel_size: int = 9):
"""Simple CNN with `n_hidden_layers`, `n_kernels`, and `kernel_size` as hyperparameters"""
super().__init__()
ksize = kernel_size
cnn = []
for i in range(n_hidden_layers):
cnn.append(torch.nn.Conv2d(
in_channels=n_in_channels,
out_channels=n_kernels,
kernel_size=kernel_size,
padding=int(kernel_size / 2)
kernel_size=ksize,
padding=int(ksize / 2),
padding_mode="replicate"
))
kernel_size -= 2
cnn.append(torch.nn.ReLU())
n_in_channels = n_kernels
self.hidden_layers = torch.nn.Sequential(*cnn)

View File

@ -36,7 +36,10 @@ def eval_evalset():
piclen = len(b['input_arrays'])
for input_array, known_array in zip(b['input_arrays'], b['known_arrays']):
input_array = DataLoader.preprocess(input_array, precision=np.float32)
input_tensor = torch.cat((torch.from_numpy(input_array), torch.from_numpy(known_array)), 0)
input_array = np.expand_dims(input_array, 0)
known_array = np.expand_dims(known_array, 0)
input_tensor = torch.cat((torch.from_numpy(input_array), torch.from_numpy(known_array)), 1)
out = model(input_tensor)
out = DataLoader.postprocess(out.cpu().detach().numpy())