| from PIL import Image | |
| import numpy as np | |
| import glob | |
| import pdb | |
| import os | |
| from tqdm import tqdm | |
| train_dir = '../data/train/label' | |
| val_dir = '../data/val/label' | |
| test_dir = '../data/test_indomain/label' | |
| dir_list = [train_dir, val_dir, test_dir] | |
| hand_cnt = 0; obj_cnt = 0 | |
| for dir in dir_list: | |
| for file in tqdm(glob.glob(dir + '/*')): | |
| lbl = np.array(Image.open(file)) | |
| unique_vals = np.unique(lbl) | |
| for unique_val in unique_vals: | |
| if unique_val == 1 or unique_val == 2: | |
| hand_cnt += 1 | |
| elif unique_val > 2: | |
| obj_cnt += 1 | |
| print(hand_cnt, obj_cnt) | |
| pdb.set_trace() | |