Shortcuts

Transfer Learning Tutorial

Author: Sasank Chilamkurthy

In this tutorial, you will learn how to train your network 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])
../_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()

    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':
                scheduler.step()
                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)

            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
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: 0.5680 Acc: 0.7008
val Loss: 0.2543 Acc: 0.8693

Epoch 1/24
----------
train Loss: 0.4543 Acc: 0.8033
val Loss: 0.1683 Acc: 0.9216

Epoch 2/24
----------
train Loss: 0.6555 Acc: 0.7869
val Loss: 0.3756 Acc: 0.8693

Epoch 3/24
----------
train Loss: 0.5446 Acc: 0.7746
val Loss: 0.2831 Acc: 0.8693

Epoch 4/24
----------
train Loss: 0.5528 Acc: 0.7869
val Loss: 0.2181 Acc: 0.9216

Epoch 5/24
----------
train Loss: 0.4406 Acc: 0.8074
val Loss: 0.2131 Acc: 0.9150

Epoch 6/24
----------
train Loss: 0.6157 Acc: 0.7746
val Loss: 0.4179 Acc: 0.8758

Epoch 7/24
----------
train Loss: 0.3954 Acc: 0.8443
val Loss: 0.2354 Acc: 0.9150

Epoch 8/24
----------
train Loss: 0.2661 Acc: 0.8811
val Loss: 0.2131 Acc: 0.9216

Epoch 9/24
----------
train Loss: 0.2670 Acc: 0.8852
val Loss: 0.2324 Acc: 0.9150

Epoch 10/24
----------
train Loss: 0.2884 Acc: 0.8852
val Loss: 0.1882 Acc: 0.9346

Epoch 11/24
----------
train Loss: 0.2641 Acc: 0.8975
val Loss: 0.1887 Acc: 0.9412

Epoch 12/24
----------
train Loss: 0.3572 Acc: 0.8607
val Loss: 0.1725 Acc: 0.9542

Epoch 13/24
----------
train Loss: 0.2219 Acc: 0.9016
val Loss: 0.1795 Acc: 0.9346

Epoch 14/24
----------
train Loss: 0.2014 Acc: 0.9262
val Loss: 0.1833 Acc: 0.9412

Epoch 15/24
----------
train Loss: 0.3065 Acc: 0.8607
val Loss: 0.1670 Acc: 0.9477

Epoch 16/24
----------
train Loss: 0.2074 Acc: 0.9016
val Loss: 0.1728 Acc: 0.9477

Epoch 17/24
----------
train Loss: 0.3417 Acc: 0.8607
val Loss: 0.2030 Acc: 0.9346

Epoch 18/24
----------
train Loss: 0.2427 Acc: 0.9221
val Loss: 0.1954 Acc: 0.9346

Epoch 19/24
----------
train Loss: 0.2520 Acc: 0.9057
val Loss: 0.1751 Acc: 0.9412

Epoch 20/24
----------
train Loss: 0.3557 Acc: 0.8156
val Loss: 0.1863 Acc: 0.9281

Epoch 21/24
----------
train Loss: 0.3184 Acc: 0.8730
val Loss: 0.1740 Acc: 0.9346

Epoch 22/24
----------
train Loss: 0.2821 Acc: 0.8648
val Loss: 0.1891 Acc: 0.9412

Epoch 23/24
----------
train Loss: 0.2105 Acc: 0.9139
val Loss: 0.1811 Acc: 0.9346

Epoch 24/24
----------
train Loss: 0.2426 Acc: 0.8975
val Loss: 0.2082 Acc: 0.9346

Training complete in 1m 13s
Best val Acc: 0.954248
visualize_model(model_ft)
../_images/sphx_glr_transfer_learning_tutorial_002.png

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: 0.5616 Acc: 0.7254
val Loss: 0.2274 Acc: 0.9150

Epoch 1/24
----------
train Loss: 0.5416 Acc: 0.7541
val Loss: 0.2052 Acc: 0.9346

Epoch 2/24
----------
train Loss: 0.4014 Acc: 0.8279
val Loss: 0.2916 Acc: 0.8889

Epoch 3/24
----------
train Loss: 0.5196 Acc: 0.7746
val Loss: 0.2076 Acc: 0.9216

Epoch 4/24
----------
train Loss: 0.4224 Acc: 0.8033
val Loss: 0.1968 Acc: 0.9477

Epoch 5/24
----------
train Loss: 0.3793 Acc: 0.8484
val Loss: 0.2387 Acc: 0.9281

Epoch 6/24
----------
train Loss: 0.4093 Acc: 0.8197
val Loss: 0.2449 Acc: 0.9281

Epoch 7/24
----------
train Loss: 0.4285 Acc: 0.8074
val Loss: 0.2174 Acc: 0.9477

Epoch 8/24
----------
train Loss: 0.3572 Acc: 0.8566
val Loss: 0.2218 Acc: 0.9477

Epoch 9/24
----------
train Loss: 0.3992 Acc: 0.8361
val Loss: 0.2251 Acc: 0.9412

Epoch 10/24
----------
train Loss: 0.2883 Acc: 0.8893
val Loss: 0.2223 Acc: 0.9346

Epoch 11/24
----------
train Loss: 0.3339 Acc: 0.8320
val Loss: 0.2216 Acc: 0.9346

Epoch 12/24
----------
train Loss: 0.3704 Acc: 0.8361
val Loss: 0.2308 Acc: 0.9412

Epoch 13/24
----------
train Loss: 0.2699 Acc: 0.8811
val Loss: 0.2381 Acc: 0.9281

Epoch 14/24
----------
train Loss: 0.2880 Acc: 0.8443
val Loss: 0.2178 Acc: 0.9412

Epoch 15/24
----------
train Loss: 0.3462 Acc: 0.8689
val Loss: 0.2458 Acc: 0.9281

Epoch 16/24
----------
train Loss: 0.2810 Acc: 0.8811
val Loss: 0.2046 Acc: 0.9477

Epoch 17/24
----------
train Loss: 0.3319 Acc: 0.8525
val Loss: 0.2217 Acc: 0.9412

Epoch 18/24
----------
train Loss: 0.3044 Acc: 0.8648
val Loss: 0.2061 Acc: 0.9412

Epoch 19/24
----------
train Loss: 0.3213 Acc: 0.8689
val Loss: 0.2357 Acc: 0.9412

Epoch 20/24
----------
train Loss: 0.3165 Acc: 0.8566
val Loss: 0.2164 Acc: 0.9412

Epoch 21/24
----------
train Loss: 0.3825 Acc: 0.8566
val Loss: 0.2521 Acc: 0.9281

Epoch 22/24
----------
train Loss: 0.2963 Acc: 0.8402
val Loss: 0.2200 Acc: 0.9477

Epoch 23/24
----------
train Loss: 0.3636 Acc: 0.8607
val Loss: 0.2235 Acc: 0.9281

Epoch 24/24
----------
train Loss: 0.3125 Acc: 0.8648
val Loss: 0.2379 Acc: 0.9281

Training complete in 0m 34s
Best val Acc: 0.947712
visualize_model(model_conv)

plt.ioff()
plt.show()
../_images/sphx_glr_transfer_learning_tutorial_003.png

Total running time of the script: ( 1 minutes 59.632 seconds)

Gallery generated by Sphinx-Gallery

Docs

Access comprehensive developer documentation for PyTorch

View Docs

Tutorials

Get in-depth tutorials for beginners and advanced developers

View Tutorials

Resources

Find development resources and get your questions answered

View Resources