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 = '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':
                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: 0.5633 Acc: 0.7090
val Loss: 0.2673 Acc: 0.9085

Epoch 1/24
----------
train Loss: 0.4639 Acc: 0.7705
val Loss: 0.2186 Acc: 0.9281

Epoch 2/24
----------
train Loss: 0.5595 Acc: 0.7787
val Loss: 0.4755 Acc: 0.7908

Epoch 3/24
----------
train Loss: 0.5109 Acc: 0.7828
val Loss: 0.2723 Acc: 0.9020

Epoch 4/24
----------
train Loss: 0.3687 Acc: 0.8402
val Loss: 0.2829 Acc: 0.8824

Epoch 5/24
----------
train Loss: 0.4498 Acc: 0.8156
val Loss: 0.4561 Acc: 0.8497

Epoch 6/24
----------
train Loss: 0.3679 Acc: 0.8525
val Loss: 0.3909 Acc: 0.8627

Epoch 7/24
----------
train Loss: 0.4990 Acc: 0.7787
val Loss: 0.3572 Acc: 0.9020

Epoch 8/24
----------
train Loss: 0.2542 Acc: 0.8975
val Loss: 0.2660 Acc: 0.9020

Epoch 9/24
----------
train Loss: 0.2219 Acc: 0.9139
val Loss: 0.3017 Acc: 0.9150

Epoch 10/24
----------
train Loss: 0.2680 Acc: 0.8975
val Loss: 0.2708 Acc: 0.9085

Epoch 11/24
----------
train Loss: 0.2899 Acc: 0.8443
val Loss: 0.2763 Acc: 0.9085

Epoch 12/24
----------
train Loss: 0.2911 Acc: 0.8730
val Loss: 0.3451 Acc: 0.8889

Epoch 13/24
----------
train Loss: 0.3719 Acc: 0.8238
val Loss: 0.2487 Acc: 0.9150

Epoch 14/24
----------
train Loss: 0.2009 Acc: 0.9139
val Loss: 0.2474 Acc: 0.9085

Epoch 15/24
----------
train Loss: 0.3487 Acc: 0.8648
val Loss: 0.2540 Acc: 0.9150

Epoch 16/24
----------
train Loss: 0.2215 Acc: 0.9016
val Loss: 0.2527 Acc: 0.9020

Epoch 17/24
----------
train Loss: 0.2705 Acc: 0.8893
val Loss: 0.2733 Acc: 0.9020

Epoch 18/24
----------
train Loss: 0.3907 Acc: 0.7992
val Loss: 0.2906 Acc: 0.8954

Epoch 19/24
----------
train Loss: 0.2485 Acc: 0.8975
val Loss: 0.2818 Acc: 0.8954

Epoch 20/24
----------
train Loss: 0.3217 Acc: 0.8648
val Loss: 0.2503 Acc: 0.9150

Epoch 21/24
----------
train Loss: 0.3598 Acc: 0.8443
val Loss: 0.2723 Acc: 0.9085

Epoch 22/24
----------
train Loss: 0.2613 Acc: 0.8852
val Loss: 0.2500 Acc: 0.9150

Epoch 23/24
----------
train Loss: 0.2734 Acc: 0.8811
val Loss: 0.2604 Acc: 0.9085

Epoch 24/24
----------
train Loss: 0.2412 Acc: 0.9057
val Loss: 0.2822 Acc: 0.8954

Training complete in 1m 15s
Best val Acc: 0.928105

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
# opoosed 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: 0.6581 Acc: 0.6680
val Loss: 0.2381 Acc: 0.9346

Epoch 1/24
----------
train Loss: 0.4257 Acc: 0.8115
val Loss: 0.2259 Acc: 0.9216

Epoch 2/24
----------
train Loss: 0.5148 Acc: 0.7623
val Loss: 0.3861 Acc: 0.8431

Epoch 3/24
----------
train Loss: 0.4335 Acc: 0.7787
val Loss: 0.1948 Acc: 0.9477

Epoch 4/24
----------
train Loss: 0.5371 Acc: 0.7500
val Loss: 0.2583 Acc: 0.9216

Epoch 5/24
----------
train Loss: 0.3293 Acc: 0.8525
val Loss: 0.1729 Acc: 0.9477

Epoch 6/24
----------
train Loss: 0.4061 Acc: 0.8156
val Loss: 0.2466 Acc: 0.9150

Epoch 7/24
----------
train Loss: 0.3684 Acc: 0.8361
val Loss: 0.1640 Acc: 0.9477

Epoch 8/24
----------
train Loss: 0.4425 Acc: 0.8033
val Loss: 0.1714 Acc: 0.9412

Epoch 9/24
----------
train Loss: 0.3291 Acc: 0.8730
val Loss: 0.1953 Acc: 0.9346

Epoch 10/24
----------
train Loss: 0.3737 Acc: 0.8074
val Loss: 0.1800 Acc: 0.9477

Epoch 11/24
----------
train Loss: 0.4401 Acc: 0.8074
val Loss: 0.1768 Acc: 0.9346

Epoch 12/24
----------
train Loss: 0.4338 Acc: 0.8115
val Loss: 0.1785 Acc: 0.9412

Epoch 13/24
----------
train Loss: 0.3561 Acc: 0.8361
val Loss: 0.1585 Acc: 0.9477

Epoch 14/24
----------
train Loss: 0.3225 Acc: 0.8811
val Loss: 0.1779 Acc: 0.9477

Epoch 15/24
----------
train Loss: 0.3239 Acc: 0.8525
val Loss: 0.1695 Acc: 0.9477

Epoch 16/24
----------
train Loss: 0.3262 Acc: 0.8648
val Loss: 0.1670 Acc: 0.9477

Epoch 17/24
----------
train Loss: 0.3496 Acc: 0.8484
val Loss: 0.1726 Acc: 0.9477

Epoch 18/24
----------
train Loss: 0.3621 Acc: 0.8361
val Loss: 0.1743 Acc: 0.9477

Epoch 19/24
----------
train Loss: 0.3848 Acc: 0.8443
val Loss: 0.1588 Acc: 0.9542

Epoch 20/24
----------
train Loss: 0.3806 Acc: 0.8279
val Loss: 0.1663 Acc: 0.9477

Epoch 21/24
----------
train Loss: 0.3252 Acc: 0.8566
val Loss: 0.1597 Acc: 0.9542

Epoch 22/24
----------
train Loss: 0.2819 Acc: 0.8607
val Loss: 0.1873 Acc: 0.9477

Epoch 23/24
----------
train Loss: 0.3235 Acc: 0.8648
val Loss: 0.1761 Acc: 0.9477

Epoch 24/24
----------
train Loss: 0.3144 Acc: 0.8525
val Loss: 0.1729 Acc: 0.9477

Training complete in 0m 38s
Best val Acc: 0.954248

visualize_model(model_conv)

plt.ioff()
plt.show()

Total running time of the script: ( 1 minutes 59.671 seconds)