Note
Go to the end to download the full example code.
Transfer Learning for Computer Vision Tutorial#
Created On: Mar 24, 2017 | Last Updated: Jan 27, 2025 | Last Verified: Nov 05, 2024
Author: Sasank Chilamkurthy
In this tutorial, you will learn how to train a convolutional neural network for image classification using transfer learning. You can read more about the transfer learning at cs231n notes
Quoting these notes,
In practice, very few people train an entire Convolutional Network from scratch (with random initialization), because it is relatively rare to have a dataset of sufficient size. Instead, it is common to pretrain a ConvNet on a very large dataset (e.g. ImageNet, which contains 1.2 million images with 1000 categories), and then use the ConvNet either as an initialization or a fixed feature extractor for the task of interest.
These two major transfer learning scenarios look as follows:
Finetuning the ConvNet: Instead of random initialization, we initialize the network with a pretrained network, like the one that is trained on imagenet 1000 dataset. Rest of the training looks as usual.
ConvNet as fixed feature extractor: Here, we will freeze the weights for all of the network except that of the final fully connected layer. This last fully connected layer is replaced with a new one with random weights and only this layer is trained.
# License: BSD
# Author: Sasank Chilamkurthy
import torch
import torch.nn as nn
import torch.optim as optim
from torch.optim import lr_scheduler
import torch.backends.cudnn as cudnn
import numpy as np
import torchvision
from torchvision import datasets, models, transforms
import matplotlib.pyplot as plt
import time
import os
from PIL import Image
from tempfile import TemporaryDirectory
cudnn.benchmark = True
plt.ion() # interactive mode
<contextlib.ExitStack object at 0x7faaf5c14190>
Load Data#
We will use torchvision and torch.utils.data packages for loading the data.
The problem we’re going to solve today is to train a model to classify ants and bees. We have about 120 training images each for ants and bees. There are 75 validation images for each class. Usually, this is a very small dataset to generalize upon, if trained from scratch. Since we are using transfer learning, we should be able to generalize reasonably well.
This dataset is a very small subset of imagenet.
Note
Download the data from here and extract it to the current directory.
# Data augmentation and normalization for training
# Just normalization for validation
data_transforms = {
'train': transforms.Compose([
transforms.RandomResizedCrop(224),
transforms.RandomHorizontalFlip(),
transforms.ToTensor(),
transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])
]),
'val': transforms.Compose([
transforms.Resize(256),
transforms.CenterCrop(224),
transforms.ToTensor(),
transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])
]),
}
data_dir = 'data/hymenoptera_data'
image_datasets = {x: datasets.ImageFolder(os.path.join(data_dir, x),
data_transforms[x])
for x in ['train', 'val']}
dataloaders = {x: torch.utils.data.DataLoader(image_datasets[x], batch_size=4,
shuffle=True, num_workers=4)
for x in ['train', 'val']}
dataset_sizes = {x: len(image_datasets[x]) for x in ['train', 'val']}
class_names = image_datasets['train'].classes
# We want to be able to train our model on an `accelerator <https://pytorch.org/docs/stable/torch.html#accelerators>`__
# such as CUDA, MPS, MTIA, or XPU. If the current accelerator is available, we will use it. Otherwise, we use the CPU.
device = torch.accelerator.current_accelerator().type if torch.accelerator.is_available() else "cpu"
print(f"Using {device} device")
Using cuda device
Visualize a few images#
Let’s visualize a few training images so as to understand the data augmentations.
def imshow(inp, title=None):
"""Display image for Tensor."""
inp = inp.numpy().transpose((1, 2, 0))
mean = np.array([0.485, 0.456, 0.406])
std = np.array([0.229, 0.224, 0.225])
inp = std * inp + mean
inp = np.clip(inp, 0, 1)
plt.imshow(inp)
if title is not None:
plt.title(title)
plt.pause(0.001) # pause a bit so that plots are updated
# Get a batch of training data
inputs, classes = next(iter(dataloaders['train']))
# Make a grid from batch
out = torchvision.utils.make_grid(inputs)
imshow(out, title=[class_names[x] for x in classes])
![['bees', 'ants', 'bees', 'bees']](../_images/sphx_glr_transfer_learning_tutorial_001.png)
Training the model#
Now, let’s write a general function to train a model. Here, we will illustrate:
Scheduling the learning rate
Saving the best model
In the following, parameter scheduler is an LR scheduler object from
torch.optim.lr_scheduler.
def train_model(model, criterion, optimizer, scheduler, num_epochs=25):
since = time.time()
# Create a temporary directory to save training checkpoints
with TemporaryDirectory() as tempdir:
best_model_params_path = os.path.join(tempdir, 'best_model_params.pt')
torch.save(model.state_dict(), best_model_params_path)
best_acc = 0.0
for epoch in range(num_epochs):
print(f'Epoch {epoch}/{num_epochs - 1}')
print('-' * 10)
# Each epoch has a training and validation phase
for phase in ['train', 'val']:
if phase == 'train':
model.train() # Set model to training mode
else:
model.eval() # Set model to evaluate mode
running_loss = 0.0
running_corrects = 0
# Iterate over data.
for inputs, labels in dataloaders[phase]:
inputs = inputs.to(device)
labels = labels.to(device)
# 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)
# backward + optimize only if in training phase
if phase == 'train':
loss.backward()
optimizer.step()
# statistics
running_loss += loss.item() * inputs.size(0)
running_corrects += torch.sum(preds == labels.data)
if phase == 'train':
scheduler.step()
epoch_loss = running_loss / dataset_sizes[phase]
epoch_acc = running_corrects.double() / dataset_sizes[phase]
print(f'{phase} Loss: {epoch_loss:.4f} Acc: {epoch_acc:.4f}')
# deep copy the model
if phase == 'val' and epoch_acc > best_acc:
best_acc = epoch_acc
torch.save(model.state_dict(), best_model_params_path)
print()
time_elapsed = time.time() - since
print(f'Training complete in {time_elapsed // 60:.0f}m {time_elapsed % 60:.0f}s')
print(f'Best val Acc: {best_acc:4f}')
# load best model weights
model.load_state_dict(torch.load(best_model_params_path, weights_only=True))
return model
Visualizing the model predictions#
Generic function to display predictions for a few images
def visualize_model(model, num_images=6):
was_training = model.training
model.eval()
images_so_far = 0
fig = plt.figure()
with torch.no_grad():
for i, (inputs, labels) in enumerate(dataloaders['val']):
inputs = inputs.to(device)
labels = labels.to(device)
outputs = model(inputs)
_, preds = torch.max(outputs, 1)
for j in range(inputs.size()[0]):
images_so_far += 1
ax = plt.subplot(num_images//2, 2, images_so_far)
ax.axis('off')
ax.set_title(f'predicted: {class_names[preds[j]]}')
imshow(inputs.cpu().data[j])
if images_so_far == num_images:
model.train(mode=was_training)
return
model.train(mode=was_training)
Finetuning the ConvNet#
Load a pretrained model and reset final fully connected layer.
model_ft = models.resnet18(weights='IMAGENET1K_V1')
num_ftrs = model_ft.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_ft.fc = nn.Linear(num_ftrs, 2)
model_ft = model_ft.to(device)
criterion = nn.CrossEntropyLoss()
# Observe that all parameters are being optimized
optimizer_ft = optim.SGD(model_ft.parameters(), lr=0.001, 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)
Downloading: "https://download.pytorch.org/models/resnet18-f37072fd.pth" to /var/lib/ci-user/.cache/torch/hub/checkpoints/resnet18-f37072fd.pth
0%| | 0.00/44.7M [00:00<?, ?B/s]
90%|█████████ | 40.2M/44.7M [00:00<00:00, 422MB/s]
100%|██████████| 44.7M/44.7M [00:00<00:00, 423MB/s]
Train and evaluate#
It should take around 15-25 min on CPU. On GPU though, it takes less than a minute.
model_ft = train_model(model_ft, criterion, optimizer_ft, exp_lr_scheduler,
num_epochs=25)
Epoch 0/24
----------
train Loss: 0.5069 Acc: 0.7459
val Loss: 0.2462 Acc: 0.9281
Epoch 1/24
----------
train Loss: 0.4709 Acc: 0.7951
val Loss: 0.3186 Acc: 0.8824
Epoch 2/24
----------
train Loss: 0.4793 Acc: 0.7869
val Loss: 0.3213 Acc: 0.8954
Epoch 3/24
----------
train Loss: 0.7054 Acc: 0.7869
val Loss: 0.4579 Acc: 0.8431
Epoch 4/24
----------
train Loss: 0.7709 Acc: 0.7336
val Loss: 0.3199 Acc: 0.8824
Epoch 5/24
----------
train Loss: 0.6012 Acc: 0.7869
val Loss: 1.1302 Acc: 0.7255
Epoch 6/24
----------
train Loss: 0.6086 Acc: 0.7787
val Loss: 0.4578 Acc: 0.8497
Epoch 7/24
----------
train Loss: 0.3450 Acc: 0.8934
val Loss: 0.3105 Acc: 0.9150
Epoch 8/24
----------
train Loss: 0.3193 Acc: 0.8648
val Loss: 0.2737 Acc: 0.9150
Epoch 9/24
----------
train Loss: 0.2744 Acc: 0.8934
val Loss: 0.2642 Acc: 0.9346
Epoch 10/24
----------
train Loss: 0.3946 Acc: 0.8197
val Loss: 0.2443 Acc: 0.9281
Epoch 11/24
----------
train Loss: 0.3528 Acc: 0.8361
val Loss: 0.2200 Acc: 0.9412
Epoch 12/24
----------
train Loss: 0.2713 Acc: 0.8811
val Loss: 0.2154 Acc: 0.9346
Epoch 13/24
----------
train Loss: 0.3270 Acc: 0.8770
val Loss: 0.2335 Acc: 0.9216
Epoch 14/24
----------
train Loss: 0.2040 Acc: 0.9139
val Loss: 0.2412 Acc: 0.9150
Epoch 15/24
----------
train Loss: 0.3084 Acc: 0.8648
val Loss: 0.2164 Acc: 0.9346
Epoch 16/24
----------
train Loss: 0.2133 Acc: 0.9180
val Loss: 0.2416 Acc: 0.9085
Epoch 17/24
----------
train Loss: 0.2977 Acc: 0.8607
val Loss: 0.2298 Acc: 0.9346
Epoch 18/24
----------
train Loss: 0.2748 Acc: 0.8730
val Loss: 0.2313 Acc: 0.9412
Epoch 19/24
----------
train Loss: 0.1987 Acc: 0.9221
val Loss: 0.2263 Acc: 0.9281
Epoch 20/24
----------
train Loss: 0.2364 Acc: 0.8852
val Loss: 0.2620 Acc: 0.8954
Epoch 21/24
----------
train Loss: 0.2260 Acc: 0.9098
val Loss: 0.2317 Acc: 0.9216
Epoch 22/24
----------
train Loss: 0.3131 Acc: 0.8770
val Loss: 0.2258 Acc: 0.9346
Epoch 23/24
----------
train Loss: 0.3112 Acc: 0.8689
val Loss: 0.2263 Acc: 0.9281
Epoch 24/24
----------
train Loss: 0.2759 Acc: 0.8811
val Loss: 0.2479 Acc: 0.9346
Training complete in 0m 35s
Best val Acc: 0.941176
visualize_model(model_ft)
ConvNet as fixed feature extractor#
Here, we need to freeze all the network except the final layer. We need
to set requires_grad = False to freeze the parameters so that the
gradients are not computed in backward().
You can read more about this in the documentation here.
model_conv = torchvision.models.resnet18(weights='IMAGENET1K_V1')
for param in model_conv.parameters():
param.requires_grad = False
# Parameters of newly constructed modules have requires_grad=True by default
num_ftrs = model_conv.fc.in_features
model_conv.fc = nn.Linear(num_ftrs, 2)
model_conv = model_conv.to(device)
criterion = nn.CrossEntropyLoss()
# Observe that only parameters of final layer are being optimized as
# opposed to before.
optimizer_conv = optim.SGD(model_conv.fc.parameters(), lr=0.001, momentum=0.9)
# Decay LR by a factor of 0.1 every 7 epochs
exp_lr_scheduler = lr_scheduler.StepLR(optimizer_conv, step_size=7, gamma=0.1)
Train and evaluate#
On CPU this will take about half the time compared to previous scenario. This is expected as gradients don’t need to be computed for most of the network. However, forward does need to be computed.
model_conv = train_model(model_conv, criterion, optimizer_conv,
exp_lr_scheduler, num_epochs=25)
Epoch 0/24
----------
train Loss: 0.6545 Acc: 0.6434
val Loss: 0.3708 Acc: 0.8105
Epoch 1/24
----------
train Loss: 0.5140 Acc: 0.7500
val Loss: 0.1585 Acc: 0.9542
Epoch 2/24
----------
train Loss: 0.3527 Acc: 0.8033
val Loss: 0.1493 Acc: 0.9608
Epoch 3/24
----------
train Loss: 0.3808 Acc: 0.8197
val Loss: 0.3457 Acc: 0.8562
Epoch 4/24
----------
train Loss: 0.5269 Acc: 0.7828
val Loss: 0.4449 Acc: 0.8235
Epoch 5/24
----------
train Loss: 0.5033 Acc: 0.8238
val Loss: 0.3730 Acc: 0.8431
Epoch 6/24
----------
train Loss: 0.5803 Acc: 0.7787
val Loss: 0.1942 Acc: 0.9281
Epoch 7/24
----------
train Loss: 0.3149 Acc: 0.8934
val Loss: 0.1746 Acc: 0.9412
Epoch 8/24
----------
train Loss: 0.3388 Acc: 0.8607
val Loss: 0.1717 Acc: 0.9346
Epoch 9/24
----------
train Loss: 0.2898 Acc: 0.8770
val Loss: 0.1772 Acc: 0.9412
Epoch 10/24
----------
train Loss: 0.3857 Acc: 0.8361
val Loss: 0.1728 Acc: 0.9346
Epoch 11/24
----------
train Loss: 0.3014 Acc: 0.8770
val Loss: 0.2103 Acc: 0.9281
Epoch 12/24
----------
train Loss: 0.3127 Acc: 0.8320
val Loss: 0.1798 Acc: 0.9346
Epoch 13/24
----------
train Loss: 0.3849 Acc: 0.8484
val Loss: 0.1723 Acc: 0.9412
Epoch 14/24
----------
train Loss: 0.4034 Acc: 0.8361
val Loss: 0.1781 Acc: 0.9346
Epoch 15/24
----------
train Loss: 0.4322 Acc: 0.8074
val Loss: 0.1658 Acc: 0.9412
Epoch 16/24
----------
train Loss: 0.3401 Acc: 0.8566
val Loss: 0.1989 Acc: 0.9346
Epoch 17/24
----------
train Loss: 0.3425 Acc: 0.8525
val Loss: 0.1980 Acc: 0.9346
Epoch 18/24
----------
train Loss: 0.2817 Acc: 0.8893
val Loss: 0.1963 Acc: 0.9346
Epoch 19/24
----------
train Loss: 0.3201 Acc: 0.8648
val Loss: 0.2092 Acc: 0.9346
Epoch 20/24
----------
train Loss: 0.3691 Acc: 0.8074
val Loss: 0.1669 Acc: 0.9346
Epoch 21/24
----------
train Loss: 0.3134 Acc: 0.8975
val Loss: 0.1772 Acc: 0.9346
Epoch 22/24
----------
train Loss: 0.4238 Acc: 0.7992
val Loss: 0.1600 Acc: 0.9412
Epoch 23/24
----------
train Loss: 0.3710 Acc: 0.8361
val Loss: 0.1965 Acc: 0.9346
Epoch 24/24
----------
train Loss: 0.3440 Acc: 0.8607
val Loss: 0.1729 Acc: 0.9477
Training complete in 0m 28s
Best val Acc: 0.960784
visualize_model(model_conv)
plt.ioff()
plt.show()
Inference on custom images#
Use the trained model to make predictions on custom images and visualize the predicted class labels along with the images.
def visualize_model_predictions(model,img_path):
was_training = model.training
model.eval()
img = Image.open(img_path)
img = data_transforms['val'](img)
img = img.unsqueeze(0)
img = img.to(device)
with torch.no_grad():
outputs = model(img)
_, preds = torch.max(outputs, 1)
ax = plt.subplot(2,2,1)
ax.axis('off')
ax.set_title(f'Predicted: {class_names[preds[0]]}')
imshow(img.cpu().data[0])
model.train(mode=was_training)
visualize_model_predictions(
model_conv,
img_path='data/hymenoptera_data/val/bees/72100438_73de9f17af.jpg'
)
plt.ioff()
plt.show()
Further Learning#
If you would like to learn more about the applications of transfer learning, checkout our Quantized Transfer Learning for Computer Vision Tutorial.
Total running time of the script: (1 minutes 5.330 seconds)
