Implementation of our paper titled "ALANET: Adaptive Latent Attention Network for Joint Video Deblurring and Interpolation" accepted to ACM-MM 2020. Please refer project page for more details. This is old version of the code. Some files may be outdated. I have also attached one checkpoint for the model.
Download data video dataset and copy it to './dataset/ ' directory.
Run the following script to generate data that can be used for training and testing.
cd data
python create_dataset.py --ffmpeg_dir <path-to-ffmpeg-dir> \
--dataset_folder <path-to-store-video-data> \
--videos_folder ./dataset
Execute run.sh bash script to train the network.
NOTE: You will have to specify parameter before running run.sh script. For more details look at run.sh in the provided codes.
@inproceedings{gupta2020alanet,
title={ALANET: Adaptive Latent Attention Network for Joint Video Deblurring and Interpolation},
author={Gupta, Akash and Aich, Abhishek and Roy-Chowdhury, Amit K},
booktitle={Proceedings of the 28th ACM International Conference on Multimedia},
pages={256--264},
year={2020}
}
Please contact the first author Akash Gupta (agupt013@ucr.edu) for any questions.