Transfer Learning for Computer Vision Tutorial

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 initializaion, 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

from __future__ import print_function, division

import torch
import torch.nn as nn
import torch.optim as optim
from torch.optim import lr_scheduler
import numpy as np
import torchvision
from torchvision import datasets, models, transforms
import matplotlib.pyplot as plt
import time
import os
import copy

plt.ion()   # interactive mode

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

device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")

Visualize a few images

Let’s visualize a few training images so as to understand the data augmentations.

def imshow(inp, title=None):
    """Imshow 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()

    best_model_wts = copy.deepcopy(model.state_dict())
    best_acc = 0.0

    for epoch in range(num_epochs):
        print('Epoch {}/{}'.format(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('{} Loss: {:.4f} Acc: {:.4f}'.format(
                phase, epoch_loss, epoch_acc))

            # deep copy the model
            if phase == 'val' and epoch_acc > best_acc:
                best_acc = epoch_acc
                best_model_wts = copy.deepcopy(model.state_dict())

        print()

    time_elapsed = time.time() - since
    print('Training complete in {:.0f}m {:.0f}s'.format(
        time_elapsed // 60, time_elapsed % 60))
    print('Best val Acc: {:4f}'.format(best_acc))

    # load best model weights
    model.load_state_dict(best_model_wts)
    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('predicted: {}'.format(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(pretrained=True)
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)

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)

Out:

Epoch 0/24
----------
train Loss: 0.5292 Acc: 0.7418
val Loss: 0.2435 Acc: 0.9150

Epoch 1/24
----------
train Loss: 0.4283 Acc: 0.7951
val Loss: 0.5986 Acc: 0.7778

Epoch 2/24
----------
train Loss: 0.4589 Acc: 0.7951
val Loss: 0.3708 Acc: 0.8889

Epoch 3/24
----------
train Loss: 0.7019 Acc: 0.7623
val Loss: 0.4007 Acc: 0.8497

Epoch 4/24
----------
train Loss: 0.5174 Acc: 0.7828
val Loss: 0.5197 Acc: 0.8105

Epoch 5/24
----------
train Loss: 0.6645 Acc: 0.7705
val Loss: 0.2458 Acc: 0.9150

Epoch 6/24
----------
train Loss: 0.8091 Acc: 0.7295
val Loss: 1.2591 Acc: 0.6667

Epoch 7/24
----------
train Loss: 0.3499 Acc: 0.8730
val Loss: 0.3899 Acc: 0.8758

Epoch 8/24
----------
train Loss: 0.4131 Acc: 0.8279
val Loss: 0.2001 Acc: 0.9346

Epoch 9/24
----------
train Loss: 0.3864 Acc: 0.8320
val Loss: 0.1812 Acc: 0.9412

Epoch 10/24
----------
train Loss: 0.3383 Acc: 0.8525
val Loss: 0.1759 Acc: 0.9346

Epoch 11/24
----------
train Loss: 0.2360 Acc: 0.8975
val Loss: 0.1710 Acc: 0.9346

Epoch 12/24
----------
train Loss: 0.2938 Acc: 0.8566
val Loss: 0.2039 Acc: 0.9085

Epoch 13/24
----------
train Loss: 0.2758 Acc: 0.8975
val Loss: 0.1686 Acc: 0.9412

Epoch 14/24
----------
train Loss: 0.2382 Acc: 0.8934
val Loss: 0.2033 Acc: 0.9150

Epoch 15/24
----------
train Loss: 0.2365 Acc: 0.9262
val Loss: 0.1866 Acc: 0.9216

Epoch 16/24
----------
train Loss: 0.2843 Acc: 0.8730
val Loss: 0.1755 Acc: 0.9346

Epoch 17/24
----------
train Loss: 0.2706 Acc: 0.8934
val Loss: 0.1827 Acc: 0.9346

Epoch 18/24
----------
train Loss: 0.2851 Acc: 0.8689
val Loss: 0.1782 Acc: 0.9412

Epoch 19/24
----------
train Loss: 0.3283 Acc: 0.8689
val Loss: 0.1908 Acc: 0.9281

Epoch 20/24
----------
train Loss: 0.3592 Acc: 0.8730
val Loss: 0.1761 Acc: 0.9412

Epoch 21/24
----------
train Loss: 0.2798 Acc: 0.8811
val Loss: 0.2299 Acc: 0.9150

Epoch 22/24
----------
train Loss: 0.2619 Acc: 0.8852
val Loss: 0.1801 Acc: 0.9346

Epoch 23/24
----------
train Loss: 0.2110 Acc: 0.9221
val Loss: 0.1910 Acc: 0.9216

Epoch 24/24
----------
train Loss: 0.3152 Acc: 0.8770
val Loss: 0.1802 Acc: 0.9216

Training complete in 1m 6s
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(pretrained=True)
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)

Out:

Epoch 0/24
----------
train Loss: 0.5708 Acc: 0.7008
val Loss: 0.2438 Acc: 0.9216

Epoch 1/24
----------
train Loss: 0.4900 Acc: 0.7746
val Loss: 0.2021 Acc: 0.9281

Epoch 2/24
----------
train Loss: 0.6654 Acc: 0.7049
val Loss: 0.1479 Acc: 0.9477

Epoch 3/24
----------
train Loss: 0.4726 Acc: 0.7910
val Loss: 0.1344 Acc: 0.9412

Epoch 4/24
----------
train Loss: 0.5337 Acc: 0.7869
val Loss: 0.1848 Acc: 0.9477

Epoch 5/24
----------
train Loss: 0.3548 Acc: 0.8566
val Loss: 0.1557 Acc: 0.9412

Epoch 6/24
----------
train Loss: 0.5880 Acc: 0.7787
val Loss: 0.1836 Acc: 0.9477

Epoch 7/24
----------
train Loss: 0.4075 Acc: 0.8238
val Loss: 0.1787 Acc: 0.9412

Epoch 8/24
----------
train Loss: 0.3954 Acc: 0.8566
val Loss: 0.1560 Acc: 0.9477

Epoch 9/24
----------
train Loss: 0.3455 Acc: 0.8607
val Loss: 0.1604 Acc: 0.9477

Epoch 10/24
----------
train Loss: 0.3392 Acc: 0.8525
val Loss: 0.1544 Acc: 0.9412

Epoch 11/24
----------
train Loss: 0.3283 Acc: 0.8607
val Loss: 0.1626 Acc: 0.9477

Epoch 12/24
----------
train Loss: 0.3476 Acc: 0.8443
val Loss: 0.1777 Acc: 0.9477

Epoch 13/24
----------
train Loss: 0.3481 Acc: 0.8156
val Loss: 0.1693 Acc: 0.9477

Epoch 14/24
----------
train Loss: 0.3442 Acc: 0.8484
val Loss: 0.1766 Acc: 0.9477

Epoch 15/24
----------
train Loss: 0.3641 Acc: 0.8525
val Loss: 0.1613 Acc: 0.9477

Epoch 16/24
----------
train Loss: 0.3654 Acc: 0.8443
val Loss: 0.1607 Acc: 0.9477

Epoch 17/24
----------
train Loss: 0.3536 Acc: 0.8566
val Loss: 0.1523 Acc: 0.9477

Epoch 18/24
----------
train Loss: 0.3742 Acc: 0.8197
val Loss: 0.1695 Acc: 0.9412

Epoch 19/24
----------
train Loss: 0.3096 Acc: 0.8525
val Loss: 0.1803 Acc: 0.9477

Epoch 20/24
----------
train Loss: 0.3834 Acc: 0.8320
val Loss: 0.1654 Acc: 0.9477

Epoch 21/24
----------
train Loss: 0.3256 Acc: 0.8730
val Loss: 0.1560 Acc: 0.9477

Epoch 22/24
----------
train Loss: 0.3141 Acc: 0.8607
val Loss: 0.2021 Acc: 0.9346

Epoch 23/24
----------
train Loss: 0.2894 Acc: 0.8770
val Loss: 0.1851 Acc: 0.9412

Epoch 24/24
----------
train Loss: 0.3573 Acc: 0.8566
val Loss: 0.1879 Acc: 0.9346

Training complete in 0m 33s
Best val Acc: 0.947712

visualize_model(model_conv)

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.