UPDATE 12/2024: A large-scale high resolution dataset for video interpolation is now available check out LAVIB
Official repository for hosting tools for creating Inter4K video interpolation dataset.
Project website: Inter4K/index.html
Paper: AdaPool: Exponential Adaptive Pooling for Information-Retaining Downsampling
- FFmpeg (https://www.ffmpeg.org/download.html)
- You will first need to download the 4K 60fps video clips (~15GB in total). Use the link https://tinyurl.com/inter4KUHD either through the browser or with
wget
on terminal:
wget https://tinyurl.com/inter4KUHD
- Clone the repository:
git clone git@github.com:alexandrosstergiou/Inter4K.git
- Unzip the folder inside the repository:
unzip Inter4K.zip -d Inter4K
- Perform a check to all files by running:
cd Inter4K/
python check.py
- Finally, create the full dataset by running:
python resolution_changes.py
! Note: By default, the program will use 16 workers to create the dataset. If you are interested on speeding-up things (and given that you have a capable CPU) you can change the number of workers by editing
resolution_chanes.py line #42
. This can also be the case that a less powerful CPU is used.
@article{stergiou2021adapool,
title={AdaPool: Exponential Adaptive Pooling for Information-Retaining Downsampling},
author={Stergiou, Alexandros and Poppe, Ronald},
booktitle={arXiv},
year={2021}
}