Preprocessing for UCF101

[greggg123]

Hi, I was wondering if there are any available resources on the preprocessing for UCF101 similar to the given data_clean.py script for SSV2?

[yztongzhan]

Hi @greggg123! There is no need to preprocess the videos of UCF101. We use decord to decode the raw videos on the fly.

[yossibiton]

@yztongzhan
Where can I get the train.csv, test.csv for UCF101 ?
Can you please provide the csv files you used (as you did for Kinetics) ?

I downloaded the original dataset and it only has these txt files :
classInd.txt
testlist01.txt
testlist02.txt
testlist03.txt
trainlist01.txt
trainlist02.txt
trainlist03.txt

[wjj-w]

Hi, have you solved this problem now and what should be done with the UCF101 dataset?

[yossibiton]

I believe this dataset has 3 train/test splits.
I don't know which split was used by this repo owners.

[wjj-w]

Thank you very much for your reply. I'm trying to randomly select one of the split, but I'm having some problems in generating the csv file ， the video is not being read correctly. Have you successfully generate the csv file based on one of the splitting methods? Can you share it? Looking forward to your reply!