Table 1 Arguments of augmentation pipeline.

From: A hybrid learning network with progressive resizing and PCA for diagnosis of cervical cancer on WSI slides

S.No.

Functions

Arguments

1

aug_func

rotate (-60, 60), shear (-3, 3), translation {‘x-axis’: (-0.2, 0.2), ‘y-axis’: (-0.2, 0.2)}, scale {‘x-axis’: (0.7, 1.3), ‘y-axis’: (0.7, 1.3)}, v_flip (1.0), h_flip (1.0)

2

canny_func

canny (alpha = (0.6, 0.9), sobel_kernel_size = (2, 8))

3

color_func

add_hue_saturation (-60, 60), channel_shuffle (1.0), hue_n_saturation (0.6, 1.4), kmeans_color (n_colors = (5, 15), grayscale (1.0)

4

sharp_func

lightness = (0.8, 1.4)), sharpen (alpha = (0.3, 0.7)

5

contrast_func

contrast (LinearContrast (0.8, 1.4)), brightness (0.4,1.5), brightness_channel ((0.6, 1.4), per_channel = 0.8)

6

green_func

green_channel (Add ((20, 90), rotate = (0, 60))

7

clahe_func

clip_limit (2, 10), gamma_contrast (1.0, 3.0), tile_grid_size (4, 20), channel_clahe

8

blue_func

blue_ channel (Add ((20, 90), rotate = (0, 60))

9

flip_func

histogram_equalization, h_flip (1.0), v_flip (1.0)

10

edge_func

edge_detect (alpha = (0, 0.4), direction = (0.0, 1.0)), directed_edge_detect (alpha = (0, 0.4)