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 0x7fb0d4fa4880>

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])

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]
 92%|#########2| 41.1M/44.7M [00:00<00:00, 430MB/s]
100%|##########| 44.7M/44.7M [00:00<00:00, 431MB/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: 0.5555 Acc: 0.7172
val Loss: 0.2248 Acc: 0.9346

Epoch 1/24
----------
train Loss: 0.5569 Acc: 0.7623
val Loss: 0.2059 Acc: 0.9150

Epoch 2/24
----------
train Loss: 0.3794 Acc: 0.8361
val Loss: 0.2652 Acc: 0.9085

Epoch 3/24
----------
train Loss: 0.5047 Acc: 0.8033
val Loss: 0.7269 Acc: 0.7386

Epoch 4/24
----------
train Loss: 0.4896 Acc: 0.8197
val Loss: 0.2691 Acc: 0.8824

Epoch 5/24
----------
train Loss: 0.6649 Acc: 0.7828
val Loss: 0.1788 Acc: 0.9542

Epoch 6/24
----------
train Loss: 0.7093 Acc: 0.7664
val Loss: 0.2629 Acc: 0.8824

Epoch 7/24
----------
train Loss: 0.4124 Acc: 0.8607
val Loss: 0.2945 Acc: 0.9020

Epoch 8/24
----------
train Loss: 0.3055 Acc: 0.8811
val Loss: 0.2641 Acc: 0.9085

Epoch 9/24
----------
train Loss: 0.3147 Acc: 0.8852
val Loss: 0.2376 Acc: 0.9085

Epoch 10/24
----------
train Loss: 0.2746 Acc: 0.8893
val Loss: 0.2657 Acc: 0.9085

Epoch 11/24
----------
train Loss: 0.3460 Acc: 0.8607
val Loss: 0.2617 Acc: 0.9150

Epoch 12/24
----------
train Loss: 0.2783 Acc: 0.8975
val Loss: 0.2773 Acc: 0.9085

Epoch 13/24
----------
train Loss: 0.2749 Acc: 0.8975
val Loss: 0.2353 Acc: 0.9150

Epoch 14/24
----------
train Loss: 0.3622 Acc: 0.8443
val Loss: 0.2741 Acc: 0.9085

Epoch 15/24
----------
train Loss: 0.2757 Acc: 0.8852
val Loss: 0.2292 Acc: 0.9216

Epoch 16/24
----------
train Loss: 0.2774 Acc: 0.8893
val Loss: 0.2442 Acc: 0.9150

Epoch 17/24
----------
train Loss: 0.2316 Acc: 0.9057
val Loss: 0.2472 Acc: 0.9150

Epoch 18/24
----------
train Loss: 0.1729 Acc: 0.9385
val Loss: 0.2617 Acc: 0.9150

Epoch 19/24
----------
train Loss: 0.3305 Acc: 0.8730
val Loss: 0.2599 Acc: 0.9020

Epoch 20/24
----------
train Loss: 0.2455 Acc: 0.9180
val Loss: 0.2343 Acc: 0.9216

Epoch 21/24
----------
train Loss: 0.2983 Acc: 0.8607
val Loss: 0.2741 Acc: 0.9085

Epoch 22/24
----------
train Loss: 0.2959 Acc: 0.8811
val Loss: 0.2341 Acc: 0.9216

Epoch 23/24
----------
train Loss: 0.3422 Acc: 0.8730
val Loss: 0.2384 Acc: 0.9085

Epoch 24/24
----------
train Loss: 0.3103 Acc: 0.8689
val Loss: 0.2307 Acc: 0.9281

Training complete in 0m 34s
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: 0.7305 Acc: 0.6352
val Loss: 0.1869 Acc: 0.9150

Epoch 1/24
----------
train Loss: 0.5002 Acc: 0.7828
val Loss: 0.2263 Acc: 0.9281

Epoch 2/24
----------
train Loss: 0.4981 Acc: 0.8033
val Loss: 0.2626 Acc: 0.9085

Epoch 3/24
----------
train Loss: 0.5071 Acc: 0.7869
val Loss: 0.3526 Acc: 0.8889

Epoch 4/24
----------
train Loss: 0.3014 Acc: 0.8770
val Loss: 0.1911 Acc: 0.9412

Epoch 5/24
----------
train Loss: 0.3081 Acc: 0.8361
val Loss: 0.1920 Acc: 0.9542

Epoch 6/24
----------
train Loss: 0.3891 Acc: 0.8361
val Loss: 0.4000 Acc: 0.8627

Epoch 7/24
----------
train Loss: 0.6116 Acc: 0.7336
val Loss: 0.1843 Acc: 0.9542

Epoch 8/24
----------
train Loss: 0.3597 Acc: 0.8607
val Loss: 0.1793 Acc: 0.9477

Epoch 9/24
----------
train Loss: 0.3758 Acc: 0.8197
val Loss: 0.1740 Acc: 0.9542

Epoch 10/24
----------
train Loss: 0.4304 Acc: 0.8238
val Loss: 0.1901 Acc: 0.9542

Epoch 11/24
----------
train Loss: 0.3309 Acc: 0.8689
val Loss: 0.1769 Acc: 0.9608

Epoch 12/24
----------
train Loss: 0.3570 Acc: 0.8443
val Loss: 0.1916 Acc: 0.9542

Epoch 13/24
----------
train Loss: 0.3031 Acc: 0.8730
val Loss: 0.2070 Acc: 0.9477

Epoch 14/24
----------
train Loss: 0.3512 Acc: 0.8484
val Loss: 0.1895 Acc: 0.9477

Epoch 15/24
----------
train Loss: 0.3269 Acc: 0.8730
val Loss: 0.2028 Acc: 0.9542

Epoch 16/24
----------
train Loss: 0.3594 Acc: 0.8402
val Loss: 0.1846 Acc: 0.9608

Epoch 17/24
----------
train Loss: 0.2602 Acc: 0.9016
val Loss: 0.1910 Acc: 0.9542

Epoch 18/24
----------
train Loss: 0.4326 Acc: 0.7951
val Loss: 0.1810 Acc: 0.9542

Epoch 19/24
----------
train Loss: 0.2477 Acc: 0.8975
val Loss: 0.2231 Acc: 0.9346

Epoch 20/24
----------
train Loss: 0.2881 Acc: 0.8770
val Loss: 0.1778 Acc: 0.9608

Epoch 21/24
----------
train Loss: 0.4013 Acc: 0.7951
val Loss: 0.2179 Acc: 0.9542

Epoch 22/24
----------
train Loss: 0.3019 Acc: 0.8648
val Loss: 0.1868 Acc: 0.9542

Epoch 23/24
----------
train Loss: 0.3794 Acc: 0.8320
val Loss: 0.1763 Acc: 0.9608

Epoch 24/24
----------
train Loss: 0.3398 Acc: 0.8361
val Loss: 0.1858 Acc: 0.9477

Training complete in 0m 27s
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.