| from nntplib import NNTPDataError | |
| from random import shuffle | |
| import numpy as np | |
| import os | |
| from PIL import Image | |
| Image.MAX_IMAGE_PIXELS = None | |
| # neg_root = '/mnt/data2/lanfz/datasets/digestpath2019/tissue-train-neg/' | |
| # neg_names = os.listdir(neg_root) | |
| # for neg_name in neg_names: | |
| # neg_path = neg_root+neg_name | |
| # neg_img = Image.open(neg_path) | |
| # mask_neg = np.zeros((neg_img.size[1],neg_img.size[0]), dtype=np.uint8) | |
| # Image.fromarray(mask_neg).save('/mnt/data2/lanfz/datasets/digestpath2019/tissue-train-neg-mask/'+neg_name[:-4]+'_mask.jpg') | |
| image_dir= '/mnt/data2/lanfz/datasets/digestpath2019/tissue-train-pos-v2/' | |
| label_dir = '/mnt/data2/lanfz/datasets/digestpath2019/tissue-train-pos-v2/' | |
| # means edg to inter, no two means edg to background | |
| label_re_dir = '/mnt/data2/lanfz/datasets/digestpath2019/tissue-train-pos-v2/' | |
| label_names = os.listdir(label_dir) | |
| print(len(label_names)) | |
| for label_name in label_names: | |
| if 'mask' in label_name: | |
| label_path = label_dir + label_name | |
| # image_path = image_dir + label_name[:-9] + '.jpg' | |
| label_img = np.array(Image.open(label_path),dtype=np.uint8) | |
| print(np.unique(label_img)) | |
| # if len(np.unique(label_img))!=1 and len(np.unique(label_img))!=2: | |
| # print(np.unique(label_img),label_name) | |
| # os.remove(label_path) | |
| # os.remove(image_path) | |
| # label_img[label_img>128] = 255 | |
| # label_img[label_img<=128] = 0 | |
| # label_img[label_img==255] = 1 | |
| # print(np.unique(label_img)) | |
| # Image.fromarray(label_img).save(label_re_dir+label_name[:-4]+'.png') |