mindspore.dataset.vision.RandAugment

class mindspore.dataset.vision.RandAugment(num_ops=2, magnitude=9, num_magnitude_bins=31, interpolation=Inter.NEAREST, fill_value=0)[source]

Apply RandAugment data augmentation method on the input image.

Refer to RandAugment: Learning Augmentation Strategies from Data .

Only support 3-channel RGB image.

Parameters

num_ops (int, optional) – Number of augmentation transformations to apply sequentially. Default: 2.
magnitude (int, optional) – Magnitude for all the transformations, must be smaller than num_magnitude_bins. Default: 9.
num_magnitude_bins (int, optional) – The number of different magnitude values, must be no less than 2. Default: 31.
interpolation (Inter, optional) –
Image interpolation method. Default: Inter.NEAREST. It can be Inter.NEAREST, Inter.BILINEAR, Inter.BICUBIC or Inter.AREA.
- Inter.NEAREST , nearest-neighbor interpolation.
- Inter.BILINEA , bilinear interpolation.
- Inter.BICUBIC , bicubic interpolation.
- Inter.AREA :, pixel area interpolation.
fill_value (Union[int, tuple[int, int, int]], optional) – Pixel fill value for the area outside the transformed image, must be in range of [0, 255]. Default: 0. If int is provided, pad all RGB channels with this value. If tuple[int, int, int] is provided, pad R, G, B channels respectively.

Raises

TypeError – If num_ops is not of type int.
ValueError – If num_ops is negative.
TypeError – If magnitude is not of type int.
ValueError – If magnitude is not positive.
TypeError – If num_magnitude_bins is not of type int.
ValueError – If num_magnitude_bins is less than 2.
TypeError – If interpolation not of type mindspore.dataset.vision.Inter .
TypeError – If fill_value is not of type int or tuple[int, int, int].
ValueError – If fill_value is not in range of [0, 255].
RuntimeError – If shape of the input image is not <H, W, C>.

Supported Platforms:: CPU

Examples

>>> import mindspore.dataset as ds
>>> import mindspore.dataset.vision as vision
>>>
>>> image_folder_dataset = ds.ImageFolderDataset("/path/to/image_folder_dataset_directory")
>>> transforms_list = [vision.Decode(), vision.RandAugment()]
>>> image_folder_dataset = image_folder_dataset.map(operations=transforms_list, input_columns=["image"])

Tutorial Examples:

Illustration of vision transforms