ImageLoader

class ImageLoader(image_dir: str, targets: numpy.ndarray, channels: int = 3, im_shape: tuple = None, batch_size: int = 32, shuffle: bool = True)

Bases: bluebird.dataloader.dataloaderbase.DataLoaderBase

__init__(image_dir: str, targets: numpy.ndarray, channels: int = 3, im_shape: tuple = None, batch_size: int = 32, shuffle: bool = True)

Initalizes the object.

Parameters

• image_dir (str) – location of base images directory

• targets (Tensor) – target data, note: order must be the same as filenames in directory

• channels (int) – number of channels of images, 1: grayscale, 3: rgb, 4: rgba, defautls to 3

• im_shape (tuple, optional) – define if you want to reshape the image, None = default image shape

• batch_size (int, optional) – length of every batch size, defaults to 32

• shuffle (bool, optional) – shuffles data if true, defaults to True

Methods

__init__(image_dir, targets[, channels, …])	Initalizes the object.
preload()	Loads the file names as Tensor

preload() → None

Loads the file names as Tensor