test_try.py

import os 
import pandas as pd 

# '/inspire/hdd/ws-f4d69b29-e0a5-44e6-bd92-acf4de9990f0/gaopeng/zhoutao-240108120126/datasets/audio-visual/vggsound/vggsound.csv' #
csv_path = '/inspire/hdd/ws-f4d69b29-e0a5-44e6-bd92-acf4de9990f0/gaopeng/public/kwang/datasets/vggsound/vggsound-caption.csv'

df = pd.read_csv(csv_path, header=None, names=['id', 'sec', 'caption', 'split']).to_dict(orient='records')

for row in df:
    start_sec = int(row['sec'])
    video_id = str(row['id'])
    # this is how our videos are named
    video_name = f'{video_id}_{start_sec:06d}'
    caption = row['caption']
    print(start_sec)
    print(video_id)
    print(caption)
    break