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 0x7f3e83f12680>
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', 'ants']](../_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]
30%|███ | 13.5M/44.7M [00:00<00:00, 128MB/s]
58%|█████▊ | 25.8M/44.7M [00:00<00:00, 122MB/s]
87%|████████▋ | 38.8M/44.7M [00:00<00:00, 128MB/s]
100%|██████████| 44.7M/44.7M [00:00<00:00, 91.5MB/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: 0.6827 Acc: 0.6557
val Loss: 0.1553 Acc: 0.9542
Epoch 1/24
----------
train Loss: 0.4175 Acc: 0.8074
val Loss: 0.2165 Acc: 0.9216
Epoch 2/24
----------
train Loss: 0.5186 Acc: 0.8033
val Loss: 0.3735 Acc: 0.8105
Epoch 3/24
----------
train Loss: 0.3668 Acc: 0.8402
val Loss: 0.3149 Acc: 0.8954
Epoch 4/24
----------
train Loss: 0.5925 Acc: 0.7295
val Loss: 0.3986 Acc: 0.8235
Epoch 5/24
----------
train Loss: 0.4540 Acc: 0.8033
val Loss: 0.2845 Acc: 0.8889
Epoch 6/24
----------
train Loss: 0.5035 Acc: 0.8197
val Loss: 0.3590 Acc: 0.8758
Epoch 7/24
----------
train Loss: 0.3760 Acc: 0.8361
val Loss: 0.2390 Acc: 0.9085
Epoch 8/24
----------
train Loss: 0.2940 Acc: 0.8566
val Loss: 0.2314 Acc: 0.9216
Epoch 9/24
----------
train Loss: 0.2798 Acc: 0.8893
val Loss: 0.2199 Acc: 0.9216
Epoch 10/24
----------
train Loss: 0.2872 Acc: 0.8975
val Loss: 0.2319 Acc: 0.8889
Epoch 11/24
----------
train Loss: 0.2714 Acc: 0.8934
val Loss: 0.2278 Acc: 0.9281
Epoch 12/24
----------
train Loss: 0.2908 Acc: 0.8648
val Loss: 0.2213 Acc: 0.9020
Epoch 13/24
----------
train Loss: 0.3901 Acc: 0.8361
val Loss: 0.2166 Acc: 0.9085
Epoch 14/24
----------
train Loss: 0.2995 Acc: 0.8811
val Loss: 0.2235 Acc: 0.8954
Epoch 15/24
----------
train Loss: 0.2275 Acc: 0.9139
val Loss: 0.2289 Acc: 0.8889
Epoch 16/24
----------
train Loss: 0.2642 Acc: 0.8689
val Loss: 0.2364 Acc: 0.9085
Epoch 17/24
----------
train Loss: 0.2212 Acc: 0.9098
val Loss: 0.2302 Acc: 0.9020
Epoch 18/24
----------
train Loss: 0.2723 Acc: 0.8811
val Loss: 0.2179 Acc: 0.9020
Epoch 19/24
----------
train Loss: 0.2699 Acc: 0.9057
val Loss: 0.2181 Acc: 0.9216
Epoch 20/24
----------
train Loss: 0.2254 Acc: 0.8934
val Loss: 0.2208 Acc: 0.9085
Epoch 21/24
----------
train Loss: 0.2597 Acc: 0.8893
val Loss: 0.2104 Acc: 0.9150
Epoch 22/24
----------
train Loss: 0.2954 Acc: 0.8811
val Loss: 0.2199 Acc: 0.9085
Epoch 23/24
----------
train Loss: 0.2099 Acc: 0.9098
val Loss: 0.2025 Acc: 0.9150
Epoch 24/24
----------
train Loss: 0.3599 Acc: 0.8525
val Loss: 0.2278 Acc: 0.9020
Training complete in 0m 33s
Best val Acc: 0.954248
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: 0.5581 Acc: 0.7336
val Loss: 0.2903 Acc: 0.9085
Epoch 1/24
----------
train Loss: 0.4256 Acc: 0.8238
val Loss: 0.2123 Acc: 0.9412
Epoch 2/24
----------
train Loss: 0.3674 Acc: 0.8566
val Loss: 0.2042 Acc: 0.9412
Epoch 3/24
----------
train Loss: 0.4415 Acc: 0.7869
val Loss: 0.1974 Acc: 0.9346
Epoch 4/24
----------
train Loss: 0.4473 Acc: 0.7951
val Loss: 0.3729 Acc: 0.8693
Epoch 5/24
----------
train Loss: 0.7147 Acc: 0.7418
val Loss: 0.5764 Acc: 0.8105
Epoch 6/24
----------
train Loss: 0.4688 Acc: 0.8279
val Loss: 0.2726 Acc: 0.9020
Epoch 7/24
----------
train Loss: 0.4625 Acc: 0.8115
val Loss: 0.1919 Acc: 0.9477
Epoch 8/24
----------
train Loss: 0.3309 Acc: 0.8607
val Loss: 0.1918 Acc: 0.9477
Epoch 9/24
----------
train Loss: 0.3351 Acc: 0.8402
val Loss: 0.1866 Acc: 0.9477
Epoch 10/24
----------
train Loss: 0.3132 Acc: 0.8852
val Loss: 0.1940 Acc: 0.9477
Epoch 11/24
----------
train Loss: 0.3170 Acc: 0.8484
val Loss: 0.1916 Acc: 0.9542
Epoch 12/24
----------
train Loss: 0.3539 Acc: 0.8443
val Loss: 0.2162 Acc: 0.9542
Epoch 13/24
----------
train Loss: 0.3029 Acc: 0.8689
val Loss: 0.1771 Acc: 0.9477
Epoch 14/24
----------
train Loss: 0.3619 Acc: 0.8156
val Loss: 0.1999 Acc: 0.9477
Epoch 15/24
----------
train Loss: 0.3153 Acc: 0.8689
val Loss: 0.2149 Acc: 0.9477
Epoch 16/24
----------
train Loss: 0.3038 Acc: 0.8893
val Loss: 0.1884 Acc: 0.9542
Epoch 17/24
----------
train Loss: 0.3283 Acc: 0.8443
val Loss: 0.1739 Acc: 0.9477
Epoch 18/24
----------
train Loss: 0.3241 Acc: 0.8770
val Loss: 0.2066 Acc: 0.9542
Epoch 19/24
----------
train Loss: 0.2899 Acc: 0.8770
val Loss: 0.1972 Acc: 0.9542
Epoch 20/24
----------
train Loss: 0.2893 Acc: 0.8852
val Loss: 0.1841 Acc: 0.9477
Epoch 21/24
----------
train Loss: 0.3195 Acc: 0.8525
val Loss: 0.1868 Acc: 0.9412
Epoch 22/24
----------
train Loss: 0.2450 Acc: 0.8934
val Loss: 0.1880 Acc: 0.9477
Epoch 23/24
----------
train Loss: 0.4439 Acc: 0.7951
val Loss: 0.1911 Acc: 0.9542
Epoch 24/24
----------
train Loss: 0.3792 Acc: 0.8074
val Loss: 0.1800 Acc: 0.9412
Training complete in 0m 27s
Best val Acc: 0.954248
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 3.093 seconds)
