Auto Augmentation

Overview

MindSpore not only allows you to customize data augmentation, but also provides an auto augmentation method to automatically perform data augmentation on images based on specific policies.

Auto augmentation can be implemented based on probability or callback parameters.

Probability-Based Auto Augmentation

MindSpore provides a series of probability-based auto augmentation APIs. You can randomly select and combine various data augmentation operations to make data augmentation more flexible.

RandomApply

The RandomApply receives a data augmentation operation list and executes the data augmentation operations in the list in sequence at a certain probability or executes none of them. The default probability is 0.5.

In the following code example, the RandomCrop and RandomColorAdjust operations are executed in sequence with a probability of 0.5 or none of them are executed.

import mindspore.dataset.vision as vision
from mindspore.dataset.transforms import RandomApply

rand_apply_list = RandomApply([vision.RandomCrop(512), vision.RandomColorAdjust()])

RandomChoice

The API receives a data augmentation operation list transforms and randomly selects a data augmentation operation to perform.

In the following code example, an operation is selected from CenterCrop and RandomCrop for execution with equal probability.

import mindspore.dataset.vision as vision
from mindspore.dataset.transforms import RandomChoice

rand_choice = RandomChoice([vision.CenterCrop(512), vision.RandomCrop(512)])

RandomSelectSubpolicy

The API receives a preset policy list, including a series of sub-policy combinations. Each sub-policy consists of several data augmentation operations executed in sequence and their execution probabilities.

First, a sub-policy is randomly selected for each image with equal probability, and then operations are performed according to the probability sequence in the sub-policy.

In the following code example, two sub-policies are preset.

Sub-policy 1 contains the RandomRotation and RandomVerticalFlipoperations, whose probabilities are 0.5 and 1.0, respectively.
Sub-policy 2 contains the RandomRotation and RandomColorAdjust operations, with the probabilities of 1.0 and 0.2, respectively.

import mindspore.dataset.vision as vision
from mindspore.dataset.vision import RandomSelectSubpolicy

policy_list = [
      [(vision.RandomRotation((45, 45)), 0.5), (vision.RandomVerticalFlip(), 1.0), (vision.RandomColorAdjust(), 0.8)],
      [(vision.RandomRotation((90, 90)), 1.0), (vision.RandomColorAdjust(), 0.2)]
      ]
policy = RandomSelectSubpolicy(policy_list)

Callback Parameter-based Auto Augmentation

The sync_wait API of MindSpore supports dynamic adjustment of the data augmentation policy by batch or epoch granularity during training. You can set blocking conditions to trigger specific data augmentation operations.

sync_wait blocks the entire data processing pipeline until sync_update triggers the customized callback function. The two APIs must be used together. Their descriptions are as follows:

sync_wait(condition_name, num_batch=1, callback=None)

This API adds a blocking condition condition_name to a dataset. When sync_update is called, the specified callback function is executed.
sync_update(condition_name, num_batch=None, data=None)

This API releases the block corresponding to condition_name and triggers the specified callback function for data.

The following demonstrates the use of automatic data augmentation based on callback parameters.

Customize the Augment class where preprocess is a custom data augmentation function and update is a callback function for updating the data augmentation policy.

import mindspore.dataset as ds
import numpy as np

class Augment:
    def __init__(self):
        self.ep_num = 0
        self.step_num = 0

    def preprocess(self, input_):
        return np.array((input_ + self.step_num ** self.ep_num - 1), )

    def update(self, data):
        self.ep_num = data['ep_num']
        self.step_num = data['step_num']

The data processing pipeline calls back the custom data augmentation policy update function update, and then performs the data augmentation operation defined in preprocess based on the updated policy in the map operation.

arr = list(range(1, 4))
dataset = ds.NumpySlicesDataset(arr, shuffle=False)
aug = Augment()
dataset = dataset.sync_wait(condition_name="policy", callback=aug.update)
dataset = dataset.map(operations=[aug.preprocess])

Call sync_update in each step to update the data augmentation policy.

epochs = 5
itr = dataset.create_tuple_iterator(num_epochs=epochs)
step_num = 0
for ep_num in range(epochs):
    for data in itr:
        print("epcoh: {}, step:{}, data :{}".format(ep_num, step_num, data))
        step_num += 1
        dataset.sync_update(condition_name="policy", data={'ep_num': ep_num, 'step_num': step_num})

The output is as follows:

epcoh: 0, step:0, data :[Tensor(shape=[], dtype=Int64, value= 1)]
epcoh: 0, step:1, data :[Tensor(shape=[], dtype=Int64, value= 2)]
epcoh: 0, step:2, data :[Tensor(shape=[], dtype=Int64, value= 3)]
epcoh: 1, step:3, data :[Tensor(shape=[], dtype=Int64, value= 1)]
epcoh: 1, step:4, data :[Tensor(shape=[], dtype=Int64, value= 5)]
epcoh: 1, step:5, data :[Tensor(shape=[], dtype=Int64, value= 7)]
epcoh: 2, step:6, data :[Tensor(shape=[], dtype=Int64, value= 6)]
epcoh: 2, step:7, data :[Tensor(shape=[], dtype=Int64, value= 50)]
epcoh: 2, step:8, data :[Tensor(shape=[], dtype=Int64, value= 66)]
epcoh: 3, step:9, data :[Tensor(shape=[], dtype=Int64, value= 81)]
epcoh: 3, step:10, data :[Tensor(shape=[], dtype=Int64, value= 1001)]
epcoh: 3, step:11, data :[Tensor(shape=[], dtype=Int64, value= 1333)]
epcoh: 4, step:12, data :[Tensor(shape=[], dtype=Int64, value= 1728)]
epcoh: 4, step:13, data :[Tensor(shape=[], dtype=Int64, value= 28562)]
epcoh: 4, step:14, data :[Tensor(shape=[], dtype=Int64, value= 38418)]

ImageNet Automatic Data Augmentation

The following is an example of implementing AutoAugment on an ImageNet dataset.

The data augmentation policy for the ImageNet dataset contains 25 sub-strategies, each of which contains two transformations. A combination of sub-strategies is randomly selected for each image in a batch, and each transformation in the sub-strategy is determined by predetermined probability.

Users can use the RandomSelectSubpolicy interface of the mindspore.dataset.vision module in MindSpore to implement AutoAugment, and the standard data augmentation method in ImageNet classification training is divided into the following steps:

RandomCropDecodeResize: Decoding after random cropping.
RandomHorizontalFlip: Flipping randomly horizontally.
Normalize: Normalization.
HWC2CHW: Changing picture channel.

Define the mapping of the MindSpore operator to the AutoAugment operator:

import mindspore.dataset.vision as vision
import mindspore.dataset.transforms as transforms

# define Auto Augmentation operators
PARAMETER_MAX = 10

def float_parameter(level, maxval):
    return float(level) * maxval /  PARAMETER_MAX

def int_parameter(level, maxval):
    return int(level * maxval / PARAMETER_MAX)

def shear_x(level):
    transforms_list = []
    v = float_parameter(level, 0.3)

    transforms_list.append(vision.RandomAffine(degrees=0, shear=(-v, -v)))
    transforms_list.append(vision.RandomAffine(degrees=0, shear=(v, v)))
    return transforms.RandomChoice(transforms_list)

def shear_y(level):
    transforms_list = []
    v = float_parameter(level, 0.3)

    transforms_list.append(vision.RandomAffine(degrees=0, shear=(0, 0, -v, -v)))
    transforms_list.append(vision.RandomAffine(degrees=0, shear=(0, 0, v, v)))
    return transforms.RandomChoice(transforms_list)

def translate_x(level):
    transforms_list = []
    v = float_parameter(level, 150 / 331)

    transforms_list.append(vision.RandomAffine(degrees=0, translate=(-v, -v)))
    transforms_list.append(vision.RandomAffine(degrees=0, translate=(v, v)))
    return transforms.RandomChoice(transforms_list)

def translate_y(level):
    transforms_list = []
    v = float_parameter(level, 150 / 331)

    transforms_list.append(vision.RandomAffine(degrees=0, translate=(0, 0, -v, -v)))
    transforms_list.append(vision.RandomAffine(degrees=0, translate=(0, 0, v, v)))
    return transforms.RandomChoice(transforms_list)

def color_impl(level):
    v = float_parameter(level, 1.8) + 0.1
    return vision.RandomColor(degrees=(v, v))

def rotate_impl(level):
    transforms_list = []
    v = int_parameter(level, 30)

    transforms_list.append(vision.RandomRotation(degrees=(-v, -v)))
    transforms_list.append(vision.RandomRotation(degrees=(v, v)))
    return transforms.RandomChoice(transforms_list)

def solarize_impl(level):
    level = int_parameter(level, 256)
    v = 256 - level
    return vision.RandomSolarize(threshold=(0, v))

def posterize_impl(level):
    level = int_parameter(level, 4)
    v = 4 - level
    return vision.RandomPosterize(bits=(v, v))

def contrast_impl(level):
    v = float_parameter(level, 1.8) + 0.1
    return vision.RandomColorAdjust(contrast=(v, v))

def autocontrast_impl(level):
    return vision.AutoContrast()

def sharpness_impl(level):
    v = float_parameter(level, 1.8) + 0.1
    return vision.RandomSharpness(degrees=(v, v))

def brightness_impl(level):
    v = float_parameter(level, 1.8) + 0.1
    return vision.RandomColorAdjust(brightness=(v, v))

Define the AutoAugment policy for the ImageNet dataset:

# define the Auto Augmentation policy
imagenet_policy = [
    [(posterize_impl(8), 0.4), (rotate_impl(9), 0.6)],
    [(solarize_impl(5), 0.6), (autocontrast_impl(5), 0.6)],
    [(vision.Equalize(), 0.8), (vision.Equalize(), 0.6)],
    [(posterize_impl(7), 0.6), (posterize_impl(6), 0.6)],

    [(vision.Equalize(), 0.4), (solarize_impl(4), 0.2)],
    [(vision.Equalize(), 0.4), (rotate_impl(8), 0.8)],
    [(solarize_impl(3), 0.6), (vision.Equalize(), 0.6)],
    [(posterize_impl(5), 0.8), (vision.Equalize(), 1.0)],
    [(rotate_impl(3), 0.2), (solarize_impl(8), 0.6)],
    [(vision.Equalize(), 0.6), (posterize_impl(6), 0.4)],

    [(rotate_impl(8), 0.8), (color_impl(0), 0.4)],
    [(rotate_impl(9), 0.4), (vision.Equalize(), 0.6)],
    [(vision.Equalize(), 0.0), (vision.Equalize(), 0.8)],
    [(vision.Invert(), 0.6), (vision.Equalize(), 1.0)],
    [(color_impl(4), 0.6), (contrast_impl(8), 1.0)],

    [(rotate_impl(8), 0.8), (color_impl(2), 1.0)],
    [(color_impl(8), 0.8), (solarize_impl(7), 0.8)],
    [(sharpness_impl(7), 0.4), (vision.Invert(), 0.6)],
    [(shear_x(5), 0.6), (vision.Equalize(), 1.0)],
    [(color_impl(0), 0.4), (vision.Equalize(), 0.6)],

    [(vision.Equalize(), 0.4), (solarize_impl(4), 0.2)],
    [(solarize_impl(5), 0.6), (autocontrast_impl(5), 0.6)],
    [(vision.Invert(), 0.6), (vision.Equalize(), 1.0)],
    [(color_impl(4), 0.6), (contrast_impl(8), 1.0)],
    [(vision.Equalize(), 0.8), (vision.Equalize(), 0.6)],
]

Insert the AutoAugment transform after the RandomCropDecodeResize operation.

import mindspore.dataset as ds
import mindspore as ms


def create_dataset(dataset_path, train, repeat_num=1,
                   batch_size=32, shuffle=False, num_samples=5):
    # create a train or eval imagenet2012 dataset for ResNet-50
    dataset = ds.ImageFolderDataset(dataset_path, num_parallel_workers=8,
                                    shuffle=shuffle, num_samples=num_samples, decode=True)

    image_size = 224

    # define map operations
    if train:
        trans = RandomSelectSubpolicy(imagenet_policy)
    else:
        trans = [vision.Resize(256),
                 vision.CenterCrop(image_size)]
    type_cast_op = transforms.TypeCast(ms.int32)

    # map images and labes
    dataset = dataset.map(operations=trans, input_columns="image")
    dataset = dataset.map(operations=type_cast_op, input_columns="label")

    # apply the batch and repeat operation
    dataset = dataset.batch(batch_size, drop_remainder=True)
    dataset = dataset.repeat(repeat_num)

    return dataset

Verify automatic data augmentations:

import matplotlib.pyplot as plt

# Define the path to image folder directory.
DATA_DIR = "/path/to/image_folder_directory"
dataset = create_dataset(dataset_path=DATA_DIR,
                         train=True,
                         batch_size=5,
                         shuffle=False,
                         num_samples=5)

epochs = 5
columns = 5
rows = 5
step_num = 0
fig = plt.figure(figsize=(8, 8))
itr = dataset.create_dict_iterator()

for ep_num in range(epochs):
    for data in itr:
        step_num += 1
        for index in range(rows):
            fig.add_subplot(rows, columns, ep_num * rows + index + 1)
            plt.imshow(data['image'].asnumpy()[index])
plt.show()

For a better demonstration of the effect, only 5 images are loaded here, and no shuffle operation is performed when reading, nor Normalize and HWC2CHW operations are performed when automatic data augmentation is performed.

augment

The running result can be seen that the augmentation effect of each image in the batch, the horizontal direction represents 5 images of 1 batch, and the vertical direction represents 5 batches.

References

[1] Ekin D. Cubuk, Barret Zoph, Dandelion Mane, Vijay Vasudevan, Quoc V. Le AutoAugment: Learning Augmentation Policies from Data.