Skip to content

Commit

Permalink
fixes
Browse files Browse the repository at this point in the history
  • Loading branch information
wassname committed Oct 13, 2020
1 parent 778c2c3 commit 5a95008
Show file tree
Hide file tree
Showing 2 changed files with 594 additions and 273 deletions.
724 changes: 535 additions & 189 deletions notebooks/c03_Finetuning/Finetuning.ipynb
View file

Large diffs are not rendered by default.

143 changes: 59 additions & 84 deletions notebooks/c03_Finetuning/Finetuning.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -115,7 +115,7 @@
#
# LANDMASS is a set of classified 2d subimages extracted from the F3 seismic volume by researchers at Georgia Tech. LANDMASS-1, contains 17667 small “patches” of size 99×99 pixels. It includes 9385 Horizon patches, 5140 chaotic patches, 1251 Fault patches, and 1891 Salt Dome patches. The images in this database have values 0-255
#
# In this notebook, we will be using the landmass dataset, which have been preprocessed already. In this dataset, we have images of 4 different types of landmass: 'Chaotic Horizon', 'Fault', 'Horizon', 'Salt Dome'.
# In this notebook, we will be using the landmass dataset, which have been preprocessed already. In this dataset, we have images of 4 different types of landmass: ['Discontinuous', 'Faulted', 'Continuous', 'Salt'].
#
# We will finetuned a pretrained CNN to learn how to classify images of landmass into those 4 classes.
#
Expand Down Expand Up @@ -147,17 +147,14 @@
transforms.ColorJitter(brightness=1, contrast=1),
transforms.ToTensor(),
transforms.Normalize((0.5,), (0.5,)),
# transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])
]
),
# For the validation dataset
"val": transforms.Compose(
[
transforms.Resize(256),
transforms.CenterCrop(224),
transforms.Resize(224),
transforms.ToTensor(),
transforms.Normalize((0.5,), (0.5,))
# transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])
]
),
}
Expand All @@ -183,18 +180,20 @@
print(landmassf3_test)
# -

# Note that this is an unbalanced dataset, so we use f1
pd.Series(landmassf3_train.train_labels).value_counts() / len(landmassf3_train)
# Note that this is an unbalanced dataset, so we use f1 score as a metric instead of accuracy
labels = pd.Series(landmassf3_train.train_labels).replace(dict(enumerate(landmassf3_train.classes)))
labels.value_counts() / len(landmassf3_train)

# +
# Our baseline f1
from sklearn.metrics import f1_score
def score_fn(y_true, y_pred):
return f1_score(y_true, y_pred, average='macro')

score_fn(
baseline_score = score_fn(
y_true=landmassf3_train.train_labels,
y_pred=[2]*len(landmassf3_train))
baseline_score
# -

classes = landmassf3_train.classes
Expand Down Expand Up @@ -245,75 +244,74 @@ def score_fn(y_true, y_pred):
def to_numpy(x):
return x.cpu().detach().numpy()

def train_model(model, dataloaders, criterion, optimizer, scheduler, num_epochs=25):
since = time.time()
dataloader = None
best_model_wts = copy.deepcopy(model.state_dict())
best_f1 = 0.0

def run_phase(model, phase, dataloader):
"""Run an epoch - through all the data once"""
if phase == "train":
model.train() # Set model to training mode
else:
model.eval() # Set model to evaluate mode
def run_phase(model, phase, dataloader, optimizer, scheduler):
"""Run an epoch - through all the data once"""
if phase == "train":
model.train() # Set model to training mode
else:
model.eval() # Set model to evaluate mode

running_loss = 0.0
running_f1 = 0
running_loss = 0.0
running_score = 0

# Iterate over data.
for inputs, labels in tqdm(dataloader, desc=phase, leave=False):
# Iterate over data.
for inputs, labels in tqdm(dataloader, desc=phase, leave=False):

# print(inputs.shape, labels.shape)
inputs = inputs.to(device)
labels = labels.to(device)
# print(inputs.shape, labels.shape)
inputs = inputs.to(device)
labels = labels.to(device)

# zero the parameter gradients
optimizer.zero_grad()

# zero the parameter gradients
optimizer.zero_grad()
# forward
# track history if only in train
with torch.set_grad_enabled(phase == "train"):
outputs = model(inputs)
_, preds = torch.max(outputs, 1)
loss = criterion(outputs, labels)

# forward
# track history if only in train
with torch.set_grad_enabled(phase == "train"):
outputs = model(inputs)
_, preds = torch.max(outputs, 1)
loss = criterion(outputs, labels)
# backward + optimize only if in training phase
if phase == "train":
loss.backward()
optimizer.step()

# backward + optimize only if in training phase
if phase == "train":
loss.backward()
optimizer.step()
# statistics
running_loss += loss.item() * inputs.size(0)
running_score += score_fn(to_numpy(labels), to_numpy(preds))
if phase == "train":
scheduler.step()

# statistics
running_loss += loss.item() * inputs.size(0)
running_f1 += score_fn(to_numpy(labels), to_numpy(preds))
if phase == "train":
scheduler.step()
epoch_loss = running_loss / len(dataloader)
epoch_score = running_score / len(dataloader)

epoch_loss = running_loss / len(dataloader)
epoch_f1 = running_f1 / len(dataloader)

return epoch_loss, epoch_f1
return epoch_loss, epoch_score

def train_model(model, dataloaders, criterion, optimizer, scheduler, num_epochs=25):
since = time.time()
dataloader = None
best_model_wts = copy.deepcopy(model.state_dict())
best_score = 0.0

try:
# Test the initial accuracy before fine tuning
epoch_loss, epoch_f1 = run_phase(model, "val", dataloaders["val"])
epoch_loss, epoch_score = run_phase(model, "val", dataloaders["val"], optimizer, scheduler)

print('Original performance without finetuning')
print("{} Loss: {:.4f} F1: {:.4f}".format("Start", epoch_loss, epoch_f1))
print("{} Loss: {:.4g} score: {:.4g}".format("Start", epoch_loss, epoch_score))

for epoch in tqdm(range(num_epochs), unit='epoch'):
print("Epoch {}/{}".format(epoch, num_epochs - 1))
print("-" * 10)

# Each epoch has a training and validation phase
for phase in ["train", "val"]:
epoch_loss, epoch_f1 = run_phase(model, phase, dataloaders[phase])
print("{} Loss: {:.4f} F1: {:.4f}".format(phase, epoch_loss, epoch_f1))
epoch_loss, epoch_score = run_phase(model, phase, dataloaders[phase], optimizer, scheduler)
print("{} Loss: {:.4g} score: {:.4g}".format(phase, epoch_loss, epoch_score))

# deep copy the model, this way we save the at the point with the best generalisation
if phase == "val" and epoch_acc > best_acc:
best_F1 = epoch_f1
if phase == "val" and epoch_score > best_score:
best_score = epoch_score
best_model_wts = copy.deepcopy(model.state_dict())

print()
Expand All @@ -326,15 +324,14 @@ def run_phase(model, phase, dataloader):
time_elapsed // 60, time_elapsed % 60
)
)
print("Best val F1: {:4f}".format(best_F1))
print("Best val score: {:4g}".format(best_score))

# load best model weights
model.load_state_dict(best_model_wts)
return model
# -



# -
# <div class="alert alert-info" style="font-size:100%">
#
# **WARNING** <br/>
Expand Down Expand Up @@ -380,14 +377,15 @@ def visualize_model(model, num_images=8, figsize=(15,15)):
return
model.train(mode=was_training)

print('Results before training')
visualize_model(model_ft, num_images=24)

# Initially the model is used to imagenet, so the intial score is very low and it gets many patches wrong. But it knows a lot about images and textures so it quickly learns.

model_ft = train_model(
model_ft, dataloaders, criterion, optimizer_ft, exp_lr_scheduler, num_epochs=2
)



visualize_model(model_ft, num_images=24)

# <div class="alert alert-success" style="font-size:100%">
Expand Down Expand Up @@ -473,31 +471,6 @@ def visualize_model(model, num_images=8, figsize=(15,15)):
#
# </details>

# +
from deep_ml_curriculum.data.landmass_f3 import LandmassF3Patches
from deep_ml_curriculum.config import project_dir

# Landmass
landmassf3_train = LandmassF3Patches(project_dir / 'data/processed/landmass-f3', train=True, transform=data_transforms['train'])
landmassf3_test = LandmassF3Patches(project_dir / 'data/processed/landmass-f3', train=False, transform=data_transforms['val'])
# Modify Resnet34 for finetuning
model_ft = models.resnet18(pretrained=False)
model_ft.conv1 = nn.Conv2d(1, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False)
model_ft.fc = nn.Linear(num_ftrs, 4)
model_ft = model_ft.to(device)

# Create Dataloader
dataloaders = {'train': DataLoader(landmassf3_train, shuffle=True, **params),
'val': DataLoader(landmassf3_test, **params)}

learning_rate = 1e-4
optimizer = optim.SGD(model_ft.parameters(), lr=learning_rate, momentum=0.9)

# Train model
model_ft = train_model(model_ft, dataloaders, criterion, optimizer, exp_lr_scheduler,
num_epochs=2)
# -

#
# # Sources
#
Expand All @@ -506,3 +479,5 @@ def visualize_model(model, num_images=8, figsize=(15,15)):





0 comments on commit 5a95008

Please sign in to comment.