beginner/transfer_learning_tutorial

Run in Google Colab

Colab

Download Notebook

Notebook

View on GitHub

GitHub

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 0x7f3fe851a470>

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]
 91%|█████████ | 40.8M/44.7M [00:00<00:00, 426MB/s]
100%|██████████| 44.7M/44.7M [00:00<00:00, 426MB/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: 0.6149 Acc: 0.6803
val Loss: 0.2027 Acc: 0.9346

Epoch 1/24
----------
train Loss: 0.4362 Acc: 0.8074
val Loss: 0.3626 Acc: 0.8431

Epoch 2/24
----------
train Loss: 0.3608 Acc: 0.8607
val Loss: 0.3606 Acc: 0.8824

Epoch 3/24
----------
train Loss: 0.6098 Acc: 0.7869
val Loss: 0.6415 Acc: 0.7843

Epoch 4/24
----------
train Loss: 0.4915 Acc: 0.8238
val Loss: 0.3797 Acc: 0.8824

Epoch 5/24
----------
train Loss: 0.5662 Acc: 0.7787
val Loss: 0.6485 Acc: 0.7843

Epoch 6/24
----------
train Loss: 0.4597 Acc: 0.8115
val Loss: 0.6148 Acc: 0.8693

Epoch 7/24
----------
train Loss: 0.5072 Acc: 0.8197
val Loss: 0.3271 Acc: 0.9020

Epoch 8/24
----------
train Loss: 0.2912 Acc: 0.8730
val Loss: 0.3280 Acc: 0.9216

Epoch 9/24
----------
train Loss: 0.2913 Acc: 0.8730
val Loss: 0.2941 Acc: 0.9346

Epoch 10/24
----------
train Loss: 0.3330 Acc: 0.8730
val Loss: 0.2897 Acc: 0.9346

Epoch 11/24
----------
train Loss: 0.2900 Acc: 0.8648
val Loss: 0.2822 Acc: 0.9281

Epoch 12/24
----------
train Loss: 0.3177 Acc: 0.8689
val Loss: 0.2717 Acc: 0.9216

Epoch 13/24
----------
train Loss: 0.2645 Acc: 0.9016
val Loss: 0.3076 Acc: 0.9216

Epoch 14/24
----------
train Loss: 0.2464 Acc: 0.8811
val Loss: 0.2830 Acc: 0.9216

Epoch 15/24
----------
train Loss: 0.2861 Acc: 0.8730
val Loss: 0.2699 Acc: 0.9216

Epoch 16/24
----------
train Loss: 0.3468 Acc: 0.8402
val Loss: 0.2777 Acc: 0.9150

Epoch 17/24
----------
train Loss: 0.2038 Acc: 0.9016
val Loss: 0.3223 Acc: 0.9085

Epoch 18/24
----------
train Loss: 0.2779 Acc: 0.9098
val Loss: 0.2821 Acc: 0.9281

Epoch 19/24
----------
train Loss: 0.3423 Acc: 0.8648
val Loss: 0.2745 Acc: 0.9216

Epoch 20/24
----------
train Loss: 0.1982 Acc: 0.9016
val Loss: 0.2782 Acc: 0.9216

Epoch 21/24
----------
train Loss: 0.2940 Acc: 0.8607
val Loss: 0.2787 Acc: 0.9150

Epoch 22/24
----------
train Loss: 0.2837 Acc: 0.9016
val Loss: 0.3217 Acc: 0.9150

Epoch 23/24
----------
train Loss: 0.2498 Acc: 0.8975
val Loss: 0.2679 Acc: 0.9216

Epoch 24/24
----------
train Loss: 0.2458 Acc: 0.9057
val Loss: 0.2577 Acc: 0.9216

Training complete in 0m 37s
Best val Acc: 0.934641

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: 0.7114 Acc: 0.6107
val Loss: 0.2578 Acc: 0.9085

Epoch 1/24
----------
train Loss: 0.3910 Acc: 0.8238
val Loss: 0.2116 Acc: 0.9216

Epoch 2/24
----------
train Loss: 0.7193 Acc: 0.6967
val Loss: 0.6624 Acc: 0.7124

Epoch 3/24
----------
train Loss: 0.5703 Acc: 0.7295
val Loss: 0.2002 Acc: 0.9477

Epoch 4/24
----------
train Loss: 0.7236 Acc: 0.7131
val Loss: 0.2422 Acc: 0.9020

Epoch 5/24
----------
train Loss: 0.6644 Acc: 0.7295
val Loss: 0.2853 Acc: 0.9085

Epoch 6/24
----------
train Loss: 0.4192 Acc: 0.8033
val Loss: 0.2167 Acc: 0.9281

Epoch 7/24
----------
train Loss: 0.3448 Acc: 0.8689
val Loss: 0.2011 Acc: 0.9346

Epoch 8/24
----------
train Loss: 0.3564 Acc: 0.8525
val Loss: 0.1700 Acc: 0.9477

Epoch 9/24
----------
train Loss: 0.3249 Acc: 0.8361
val Loss: 0.1979 Acc: 0.9346

Epoch 10/24
----------
train Loss: 0.3264 Acc: 0.8484
val Loss: 0.2059 Acc: 0.9346

Epoch 11/24
----------
train Loss: 0.3421 Acc: 0.8402
val Loss: 0.2026 Acc: 0.9281

Epoch 12/24
----------
train Loss: 0.3387 Acc: 0.8484
val Loss: 0.1806 Acc: 0.9542

Epoch 13/24
----------
train Loss: 0.3261 Acc: 0.8484
val Loss: 0.1679 Acc: 0.9608

Epoch 14/24
----------
train Loss: 0.3454 Acc: 0.8279
val Loss: 0.2037 Acc: 0.9346

Epoch 15/24
----------
train Loss: 0.3715 Acc: 0.8525
val Loss: 0.1829 Acc: 0.9542

Epoch 16/24
----------
train Loss: 0.3668 Acc: 0.8361
val Loss: 0.2125 Acc: 0.9281

Epoch 17/24
----------
train Loss: 0.3324 Acc: 0.8525
val Loss: 0.1916 Acc: 0.9281

Epoch 18/24
----------
train Loss: 0.2893 Acc: 0.8607
val Loss: 0.1829 Acc: 0.9477

Epoch 19/24
----------
train Loss: 0.3552 Acc: 0.8238
val Loss: 0.1963 Acc: 0.9477

Epoch 20/24
----------
train Loss: 0.3543 Acc: 0.8484
val Loss: 0.2073 Acc: 0.9216

Epoch 21/24
----------
train Loss: 0.3446 Acc: 0.8484
val Loss: 0.2109 Acc: 0.9346

Epoch 22/24
----------
train Loss: 0.2465 Acc: 0.8975
val Loss: 0.1841 Acc: 0.9412

Epoch 23/24
----------
train Loss: 0.2649 Acc: 0.8689
val Loss: 0.1736 Acc: 0.9477

Epoch 24/24
----------
train Loss: 0.3048 Acc: 0.8525
val Loss: 0.1857 Acc: 0.9412

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.