Datasets:
Maybe the total number of the dataset is less than 200k?
Dear author, I found that in the vggsound.csv, the train split contains 186733-2762=183971 clips, and the test split contains 15496 clips, the total number of the dataset is 183971+15496=199467 clips.(LESS THAN 200K)
Is it right? (I found this problem because I found that the vggsound.csv contains 199467 lines, and if we search ever line of it, some lines contain the word "train" (which means the railway of other objects).
Here are the samples of the Extra 'train' occurrences:
Extra 'train' occurrences: 2762
Extra 'test' occurrences: 0
Lines with extra 'train' occurrences: 2762
Sample lines with extra 'train':
['-1oRfe0FQI8', '60', 'railroad car, train wagon', 'train']
['-2-wdcN5vOw', '17', 'train whistling', 'test']
['-2JalGJn_4Y', '263', 'train whistling', 'train']
['-40DA8RPuq0', '34', 'train horning', 'train']
['-40DA8RPuq0', '216', 'train horning', 'train']
['-4ItJ9yTz_c', '42', 'train horning', 'train']
['-6ZEGCtBKqs', '226', 'railroad car, train wagon', 'train']
['-79qo5MUYBk', '207', 'train whistling', 'test']
['-9I17YJGFug', '0', 'train horning', 'train']
['-AioliAg12U', '10', 'railroad car, train wagon', 'train']
Lines with extra 'test' occurrences: 0
The dataset is really good, I just want to know the exact number of the dataset ,which will help my study. Thank you!
and PAOL8UF9UDY_000104.mp4 is broken
Hi, I guess you may need to separate the train set and test set directly from the .csv file.
If so, you could use the pandas package to separate according to the label of the fourth column.
Here is some sample code:
df = pd.read_csv('vggsound.csv')
filtered_df = df[df[3] == 'train']