import pandas as pd
import os

img_root = '/home/mly/data/datasets/text_recognition/CDLA/syn/'
train_img_root = os.path.join(img_root, 'train', 'img')
val_img_root = os.path.join(img_root, 'val', 'img')
gen_root = '/home/mly/data/datasets/text_recognition/CDLA/syn/'

train_img_list = sorted(os.listdir(train_img_root))
train_df = pd.DataFrame(columns=['path'])
train_df.path = train_img_list

val_img_list = sorted(os.listdir(val_img_root))
val_df = pd.DataFrame(columns=['path'])
val_df.path = val_img_list

train_df.to_csv(os.path.join(gen_root, 'train.csv'))
val_df.to_csv(os.path.join(gen_root, 'val.csv'))
print(f'saved to {gen_root}-train.csv/val.csv')