.. note:: :class: sphx-glr-download-link-note Click :ref:`here ` to download the full example code .. rst-class:: sphx-glr-example-title .. _sphx_glr_recipes_recipes_custom_dataset_transforms_loader.py: Developing Custom PyTorch Dataloaders ===================================== A significant amount of the effort applied to developing machine learning algorithms is related to data preparation. PyTorch provides many tools to make data loading easy and hopefully, makes your code more readable. In this recipe, you will learn how to: 1. Create a custom dataset leveraging the PyTorch dataset APIs; 2. Create callable custom transforms that can be composable; and 3. Put these components together to create a custom dataloader. Please note, to run this tutorial, ensure the following packages are installed: - ``scikit-image``: For image io and transforms - ``pandas``: For easier csv parsing As a point of attribution, this recipe is based on the original tutorial from `Sasank Chilamkurthy `__ and was later edited by `Joe Spisak `__. Setup ---------------------- First let’s import all of the needed libraries for this recipe. .. code-block:: default from __future__ import print_function, division import os import torch import pandas as pd from skimage import io, transform import numpy as np import matplotlib.pyplot as plt from torch.utils.data import Dataset, DataLoader from torchvision import transforms, utils # Ignore warnings import warnings warnings.filterwarnings("ignore") plt.ion() # interactive mode Part 1: The Dataset ------------------- The dataset we are going to deal with is that of facial pose. Overall, 68 different landmark points are annotated for each face. As a next step, please download the dataset from `here `_ so that the images are in a directory named ‘data/faces/’. **Note:** This dataset was actually generated by applying `dlib's pose estimation `_ on images from the imagenet dataset containing the ‘face’ tag. :: !wget https://download.pytorch.org/tutorial/faces.zip !mkdir data/faces/ import zipfile with zipfile.ZipFile("faces.zip","r") as zip_ref: zip_ref.extractall("/data/faces/") %cd /data/faces/ The dataset comes with a csv file with annotations which looks like this: :: image_name,part_0_x,part_0_y,part_1_x,part_1_y,part_2_x, ... ,part_67_x,part_67_y 0805personali01.jpg,27,83,27,98, ... 84,134 1084239450_e76e00b7e7.jpg,70,236,71,257, ... ,128,312 Let’s quickly read the CSV and get the annotations in an (N, 2) array where N is the number of landmarks. .. code-block:: default landmarks_frame = pd.read_csv('faces/face_landmarks.csv') n = 65 img_name = landmarks_frame.iloc[n, 0] landmarks = landmarks_frame.iloc[n, 1:] landmarks = np.asarray(landmarks) landmarks = landmarks.astype('float').reshape(-1, 2) print('Image name: {}'.format(img_name)) print('Landmarks shape: {}'.format(landmarks.shape)) print('First 4 Landmarks: {}'.format(landmarks[:4])) 1.1 Write a simple helper function to show an image ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Next let’s write a simple helper function to show an image, its landmarks and use it to show a sample. .. code-block:: default def show_landmarks(image, landmarks): """Show image with landmarks""" plt.imshow(image) plt.scatter(landmarks[:, 0], landmarks[:, 1], s=10, marker='.', c='r') plt.pause(0.001) # pause a bit so that plots are updated plt.figure() show_landmarks(io.imread(os.path.join('faces/', img_name)), landmarks) plt.show() 1.2 Create a dataset class ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Now lets talk about the PyTorch dataset class ``torch.utils.data.Dataset`` is an abstract class representing a dataset. Your custom dataset should inherit ``Dataset`` and override the following methods: - ``__len__`` so that ``len(dataset)`` returns the size of the dataset. - ``__getitem__`` to support indexing such that ``dataset[i]`` can be used to get :math:``i`` th sample Let’s create a dataset class for our face landmarks dataset. We will read the csv in ``__init__`` but leave the reading of images to ``__getitem__``. This is memory efficient because all the images are not stored in the memory at once but read as required. Here we show a sample of our dataset in the forma of a dict ``{'image': image, 'landmarks': landmarks}``. Our dataset will take an optional argument ``transform`` so that any required processing can be applied on the sample. We will see the usefulness of ``transform`` in another recipe. .. code-block:: default class FaceLandmarksDataset(Dataset): """Face Landmarks dataset.""" def __init__(self, csv_file, root_dir, transform=None): """ Args: csv_file (string): Path to the csv file with annotations. root_dir (string): Directory with all the images. transform (callable, optional): Optional transform to be applied on a sample. """ self.landmarks_frame = pd.read_csv(csv_file) self.root_dir = root_dir self.transform = transform def __len__(self): return len(self.landmarks_frame) def __getitem__(self, idx): if torch.is_tensor(idx): idx = idx.tolist() img_name = os.path.join(self.root_dir, self.landmarks_frame.iloc[idx, 0]) image = io.imread(img_name) landmarks = self.landmarks_frame.iloc[idx, 1:] landmarks = np.array([landmarks]) landmarks = landmarks.astype('float').reshape(-1, 2) sample = {'image': image, 'landmarks': landmarks} if self.transform: sample = self.transform(sample) return sample 1.3 Iterate through data samples ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Next let’s instantiate this class and iterate through the data samples. We will print the sizes of first 4 samples and show their landmarks. .. code-block:: default face_dataset = FaceLandmarksDataset(csv_file='faces/face_landmarks.csv', root_dir='faces/') fig = plt.figure() for i in range(len(face_dataset)): sample = face_dataset[i] print(i, sample['image'].shape, sample['landmarks'].shape) ax = plt.subplot(1, 4, i + 1) plt.tight_layout() ax.set_title('Sample #{}'.format(i)) ax.axis('off') show_landmarks(**sample) if i == 3: plt.show() break Part 2: Data Tranformations --------------------------- Now that we have a dataset to work with and have done some level of customization, we can move to creating custom transformations. In computer vision, these come in handy to help generalize algorithms and improve accuracy. A suite of transformations used at training time is typically referred to as data augmentation and is a common practice for modern model development. One issue common in handling datasets is that the samples may not all be the same size. Most neural networks expect the images of a fixed size. Therefore, we will need to write some prepocessing code. Let’s create three transforms: - ``Rescale``: to scale the image - ``RandomCrop``: to crop from image randomly. This is data augmentation. - ``ToTensor``: to convert the numpy images to torch images (we need to swap axes). We will write them as callable classes instead of simple functions so that parameters of the transform need not be passed everytime it’s called. For this, we just need to implement ``__call__`` method and if required, ``__init__`` method. We can then use a transform like this: :: tsfm = Transform(params) transformed_sample = tsfm(sample) Observe below how these transforms had to be applied both on the image and landmarks. 2.1 Create callable classes ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Let’s start with creating callable classes for each transform .. code-block:: default class Rescale(object): """Rescale the image in a sample to a given size. Args: output_size (tuple or int): Desired output size. If tuple, output is matched to output_size. If int, smaller of image edges is matched to output_size keeping aspect ratio the same. """ def __init__(self, output_size): assert isinstance(output_size, (int, tuple)) self.output_size = output_size def __call__(self, sample): image, landmarks = sample['image'], sample['landmarks'] h, w = image.shape[:2] if isinstance(self.output_size, int): if h > w: new_h, new_w = self.output_size * h / w, self.output_size else: new_h, new_w = self.output_size, self.output_size * w / h else: new_h, new_w = self.output_size new_h, new_w = int(new_h), int(new_w) img = transform.resize(image, (new_h, new_w)) # h and w are swapped for landmarks because for images, # x and y axes are axis 1 and 0 respectively landmarks = landmarks * [new_w / w, new_h / h] return {'image': img, 'landmarks': landmarks} class RandomCrop(object): """Crop randomly the image in a sample. Args: output_size (tuple or int): Desired output size. If int, square crop is made. """ def __init__(self, output_size): assert isinstance(output_size, (int, tuple)) if isinstance(output_size, int): self.output_size = (output_size, output_size) else: assert len(output_size) == 2 self.output_size = output_size def __call__(self, sample): image, landmarks = sample['image'], sample['landmarks'] h, w = image.shape[:2] new_h, new_w = self.output_size top = np.random.randint(0, h - new_h) left = np.random.randint(0, w - new_w) image = image[top: top + new_h, left: left + new_w] landmarks = landmarks - [left, top] return {'image': image, 'landmarks': landmarks} class ToTensor(object): """Convert ndarrays in sample to Tensors.""" def __call__(self, sample): image, landmarks = sample['image'], sample['landmarks'] # swap color axis because # numpy image: H x W x C # torch image: C X H X W image = image.transpose((2, 0, 1)) return {'image': torch.from_numpy(image), 'landmarks': torch.from_numpy(landmarks)} 2.2 Compose transforms and apply to a sample ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Next let’s compose these transforms and apply to a sample Let’s say we want to rescale the shorter side of the image to 256 and then randomly crop a square of size 224 from it. i.e, we want to compose ``Rescale`` and ``RandomCrop`` transforms. ``torchvision.transforms.Compose`` is a simple callable class which allows us to do this. .. code-block:: default scale = Rescale(256) crop = RandomCrop(128) composed = transforms.Compose([Rescale(256), RandomCrop(224)]) # Apply each of the above transforms on sample. fig = plt.figure() sample = face_dataset[65] for i, tsfrm in enumerate([scale, crop, composed]): transformed_sample = tsfrm(sample) ax = plt.subplot(1, 3, i + 1) plt.tight_layout() ax.set_title(type(tsfrm).__name__) show_landmarks(**transformed_sample) plt.show() 2.3 Iterate through the dataset ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Next we will iterate through the dataset Let’s put this all together to create a dataset with composed transforms. To summarize, every time this dataset is sampled: - An image is read from the file on the fly - Transforms are applied on the read image - Since one of the transforms is random, data is augmentated on sampling We can iterate over the created dataset with a ``for i in range`` loop as before. .. code-block:: default transformed_dataset = FaceLandmarksDataset(csv_file='faces/face_landmarks.csv', root_dir='faces/', transform=transforms.Compose([ Rescale(256), RandomCrop(224), ToTensor() ])) for i in range(len(transformed_dataset)): sample = transformed_dataset[i] print(i, sample['image'].size(), sample['landmarks'].size()) if i == 3: break Part 3: The Dataloader ---------------------- By operating on the dataset directly, we are losing out on a lot of features by using a simple ``for`` loop to iterate over the data. In particular, we are missing out on: - Batching the data - Shuffling the data - Load the data in parallel using ``multiprocessing`` workers. ``torch.utils.data.DataLoader`` is an iterator which provides all these features. Parameters used below should be clear. One parameter of interest is ``collate_fn``. You can specify how exactly the samples need to be batched using ``collate_fn``. However, default collate should work fine for most use cases. .. code-block:: default dataloader = DataLoader(transformed_dataset, batch_size=4, shuffle=True, num_workers=4) # Helper function to show a batch def show_landmarks_batch(sample_batched): """Show image with landmarks for a batch of samples.""" images_batch, landmarks_batch = \ sample_batched['image'], sample_batched['landmarks'] batch_size = len(images_batch) im_size = images_batch.size(2) grid = utils.make_grid(images_batch) plt.imshow(grid.numpy().transpose((1, 2, 0))) for i in range(batch_size): plt.scatter(landmarks_batch[i, :, 0].numpy() + i * im_size, landmarks_batch[i, :, 1].numpy(), s=10, marker='.', c='r') plt.title('Batch from dataloader') for i_batch, sample_batched in enumerate(dataloader): print(i_batch, sample_batched['image'].size(), sample_batched['landmarks'].size()) # observe 4th batch and stop. if i_batch == 3: plt.figure() show_landmarks_batch(sample_batched) plt.axis('off') plt.ioff() plt.show() break Now that you’ve learned how to create a custom dataloader with PyTorch, we recommend diving deeper into the docs and customizing your workflow even further. You can learn more in the ``torch.utils.data`` docs `here `__. .. rst-class:: sphx-glr-timing **Total running time of the script:** ( 0 minutes 0.000 seconds) .. _sphx_glr_download_recipes_recipes_custom_dataset_transforms_loader.py: .. only :: html .. container:: sphx-glr-footer :class: sphx-glr-footer-example .. container:: sphx-glr-download :download:`Download Python source code: custom_dataset_transforms_loader.py ` .. container:: sphx-glr-download :download:`Download Jupyter notebook: custom_dataset_transforms_loader.ipynb ` .. only:: html .. rst-class:: sphx-glr-signature `Gallery generated by Sphinx-Gallery `_