jaclearn.imageaug package¶

Submodules¶

jaclearn.imageaug.cblk module¶

jaclearn.imageaug.cblk.fbaug(img, target_shape=(224, 224), is_training=True, area_range=0.08, aspect_ratio=(0.75, 1.3333333333333333))[source]¶

jaclearn.imageaug.executor module¶

class jaclearn.imageaug.executor.AugmentorExecutorBase(*, random_order=False)[source]¶

Bases: object

f(augmentor, **kwargs)[source]¶

class jaclearn.imageaug.executor.ImageAugmentorExecutor(*, random_order=False)[source]¶: Bases: jaclearn.imageaug.executor.AugmentorExecutorBase

class jaclearn.imageaug.executor.ImageCoordAugmentorExecutor(*, random_order=False)[source]¶: Bases: jaclearn.imageaug.executor.AugmentorExecutorBase

jaclearn.imageaug.photography module¶

jaclearn.imageaug.photography.grayscale_augment(img, prob=0.5)[source]¶

jaclearn.imageaug.photography.brightness_augment(img, val)[source]¶

jaclearn.imageaug.photography.contrast_augment(img, val)[source]¶

jaclearn.imageaug.photography.saturation_augment(img, val)[source]¶

jaclearn.imageaug.photography.color_augment_pack(img, brightness, contrast, saturation)[source]¶

jaclearn.imageaug.photography.lighting_augment(img, std, eigval=None, eigvec=None)[source]¶

jaclearn.imageaug.shape module¶

jaclearn.imageaug.shape.random_crop(img, target_shape)[source]¶: random crop a image. output size is target_shape

jaclearn.imageaug.shape.random_crop_random_shape(img, max_shape, min_shape=0)[source]¶

jaclearn.imageaug.shape.random_crop_and_resize(img, max_shape, target_shape, min_shape=0)[source]¶

jaclearn.imageaug.shape.random_size_crop(img, target_shape, area_range, aspect_ratio=None, contiguous_ar=False, *, nr_trial=10)[source]¶: random size crop used for Facebook ImageNet data augmentation see https://github.com/facebook/fb.resnet.torch/blob/master/datasets/imagenet.lua

jaclearn.imageaug.shape.horizontal_flip_augment(img, prob)[source]¶