dataset_fixing.py

import os
import shutil
import random

dataset_path= './chest_xray/'
new_dataset_path = './chest_xray_new/'

if not os.path.exists(new_dataset_path):
   for split in ['train', 'val', 'test']:
       for cls in ['NORMAL', 'PNEUMONIA']:
           os.makedirs(f'{new_dataset_path}/{split}/{cls}', exist_ok=True)

   for cls in ['NORMAL', 'PNEUMONIA']:
       all_files = []
       for split in ['train', 'val', 'test']:
           source_folder = f'{dataset_path}/{split}/{cls}'
           files = os.listdir(source_folder)
           all_files.extend([(file, source_folder) for file in files])

       random.shuffle(all_files)

       train_files = all_files[:int(len(all_files)*0.7)]
       val_files = all_files[int(len(all_files)*0.7):int(len(all_files)*0.8)]
       test_files = all_files[int(len(all_files)*0.8):]

       for file, source_folder in train_files:
           dest = f'{new_dataset_path}/train/{cls}/{file}'
           shutil.copy(f'{source_folder}/{file}', dest)

       for file, source_folder in val_files:
           dest = f'{new_dataset_path}/val/{cls}/{file}'
           shutil.copy(f'{source_folder}/{file}', dest)

       for file, source_folder in test_files:
           dest = f'{new_dataset_path}/test/{cls}/{file}'
           shutil.copy(f'{source_folder}/{file}', dest)