preprocessing_scripts/preprocess_kumar.py

import os
from scipy.io import loadmat
import numpy as np
from PIL import Image
from tqdm import tqdm

data_directory = "/vol/data/histo_datasets/Kumar/"
dirs = ["train/", "test_same/", "test_diff/"]
for d in dirs:
    dir_list = os.listdir(data_directory + d + "Labels/")
    for i in dir_list:
        masks = loadmat(data_directory + d + "Labels/" + i)["inst_map"]
        os.makedirs(data_directory + d + "Labels_png_new/", exist_ok=True)
        Image.fromarray(masks).save(data_directory + d + "Labels_png_new/" + i[:-4] + ".png")