preprocess_imagenette.py

import os
from PIL import Image
from tqdm import tqdm
import cv2

data_fldf_path = os.path.join(os.getcwd(), 'data/imagenette2-320')
train_path = os.path.join(data_fldf_path, 'train')
test_path = os.path.join(data_fldf_path, 'val')

train_folders = os.listdir(train_path)
test_folders = os.listdir(test_path)

'''

for folder in train_folders:
    files_path = os.path.join(train_path, folder)
    image_files = os.listdir(files_path)
    for f in image_files:
        fp = os.path.join(files_path, f)
        new_path = os.path.join(train_path, f)
        os.rename(fp, new_path)
        
for folder in test_folders:
    files_path = os.path.join(test_path, folder)
    image_files = os.listdir(files_path)
    for f in image_files:
        fp = os.path.join(files_path, f)
        new_path = os.path.join(test_path, f)
        os.rename(fp, new_path)

'''

train_imgs = os.listdir(train_path)
test_imgs = os.listdir(test_path)
print('Start processing images...\n')
'''
for fl in tqdm(train_imgs, desc = 'Processing image folder'):
    #print(fl)
    fp = os.path.join(train_path, fl)
    im = cv2.imread(fp)
    im = cv2.resize(im, (300,300))
    cv2.imwrite(fp, im)
    del im
''' 

for fl in tqdm(test_imgs, desc = 'Processing image folder'):
    fp = os.path.join(test_path, fl)
    im = cv2.imread(fp)
    im = cv2.resize(im, (300,300))
    cv2.imwrite(fp, im)
    del im