Image_Augmentation_using_Augmentor

import os

try:
    import Augmentor
except ImportError:
    os.system('pip install Augmentor')

PATH="/path/to/image/folder"                  # path for images to be augmented 
n=200                                    # no of images after augmentation

p = Augmentor.Pipeline(PATH)

p.scale(probability=0.2, scale_factor=1.2)

p.resize(probability=1.0, width=120, height=120)
p.resize(probability=0.1, width=100, height=200, resample_filter=u'BICUBIC')

p.zoom(probability=0.5, min_factor=1.1, max_factor=1.5)
p.zoom_random(probability=0.5, percentage_area=0.8)

p.flip_top_bottom(probability=0.5)
p.flip_left_right(probability=0.5)

p.rotate(probability=0.7, max_left_rotation=10, max_right_rotation=10)
p.rotate90(probability=0.5)
p.rotate180(probability=0.5)
p.rotate270(probability=0.5)
p.rotate_random_90(probability=0.1)
p.rotate_without_crop(probability=0.01, max_left_rotation=20, max_right_rotation=20, expand=False)

p.shear(probability=0.2, max_shear_left=2, max_shear_right=3)

p.skew(probability=0.3, magnitude=1)
p.skew_corner(probability=0.01, magnitude=1)
p.skew_left_right(probability=0.02, magnitude=1)
p.skew_tilt(probability=0.03, magnitude=1)
p.skew_top_bottom(probability=0.04, magnitude=1)



p.random_brightness(probability=0.4, min_factor=0, max_factor=0.65)
p.random_color(probability=0.3, min_factor=0.4, max_factor=0.6)
p.random_contrast(probability=0.2, min_factor=0, max_factor=1)
# p.random_erasing(probability=0.3, rectangle_area=0.2)

# p.random_distortion(probability=1, grid_width=4, grid_height=4, magnitude=8)
# p.gaussian_distortion(probability, grid_width, grid_height, magnitude, corner, method, mex=0.5, mey=0.5, sdx=0.05, sdy=0.05)

# p.black_and_white(probability=0.6, threshold=128)
# p.greyscale(probability)

# p.histogram_equalisation(probability=1.0)
# p.invert(probability)

# p.crop_by_size(probability, width, height, centre=True)
# p.crop_centre(probability, percentage_area, randomise_percentage_area=False)
# p.crop_random(probability, percentage_area, randomise_percentage_area=False)


p.sample(n)

Check into Augmentor library for implementation of each augmentation techniques.