preprocessing_scripts/preprocess_cpm.py

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

data_directory = "/vol/data/histo_datasets/CPM_15_and_17/"
dir_list = ["cpm15/", "cpm17/test/", "cpm17/train/"]
for d in dir_list:
    mask_list = os.listdir(data_directory + d + "Labels/")
    for i in range(len(mask_list)):
        mask = loadmat(data_directory + d + "Labels/" + mask_list[i])["inst_map"]
        mask = mask.astype(np.int32)
        os.makedirs(data_directory + d + "Labels_png/", exist_ok=True)
        Image.fromarray(mask).save(data_directory + d + "Labels_png/" + mask_list[i][:-4] + ".png")