ai_lab is a library for loading datasets, data augmentation, training management, experiment management and result documentation
aug = Augment2D(
blur=0.1,
pixel_shift=0.1,
rotate_hard=True,
rotate_soft=False,
noise=0.01,
zoom=0.2,
lab_shift=0.5,
flip=True,
contrast=0.5)
data_in,labels_in = [im for _ in range(aug_count)], [label for _ in range(aug_count)] # number of samples to be dublicated and augmented in different ways
data_in,labels_in = aug.cut_size_hard(data_in,labels_in) # Randomize region/aspectratio of samples
data,labels = aug.augment(data_in,labels_in) # augment with init parameters