FAIR-Play Dataset from 2.5D Visual Sound

[Project Page] [arXiv] [Video]

2.5D Visual Sound
Ruohan Gao¹ and Kristen Grauman^2
1UT Austin, ²Facebook AI Research
In Conference on Computer Vision and Pattern Recognition (CVPR), 2019

This repository (~100G) contains the FAIR-Play dataset we collected and used in our CVPR 2019 paper. It contains 1,871 video clips and their corresponding binaural audio clips recorded in a music room. The video clip and binaural clip of the same index are roughly aligned. The splits directory contains the 10 random splits used in the paper. See PseudoBinaural for 5 more challenging splits, where there are no or less scene overlap in the training and testing splits. The code is shared at 2.5D Visual Sound Code.

Dataset Download

1. The dataset can be downloaded by cloning the repository uisng git lfs:

brew install git-lfs
git lfs clone git@github.com:facebookresearch/FAIR-Play.git
git lfs install
git lfs pull

2. If you have trouble in downloading the dataset through GitHub, you can also download it using the following commands:

wget http://dl.fbaipublicfiles.com/FAIR-Play/videos.tar.gz
wget http://dl.fbaipublicfiles.com/FAIR-Play/audios.tar.gz
wget http://dl.fbaipublicfiles.com/FAIR-Play/splits.tar.gz

3. The dataset is also shared at UT Box.

If you find our data or project useful in your research, please cite:

    @inproceedings{gao2019visualsound,
      title={2.5D Visual Sound},
      author={Gao, Ruohan and Grauman, Kristen},
      booktitle={CVPR},
      year={2019}
    }

Acknowlegements

We would like to thank Tony Miller, Jacob Donley, Pablo Hoffmann and Vladimir Tourbabin from Facebook for helpful discussions and the volunteers who participate in our data collection.

Licence

FAIR-Play is CC BY 4.0 licensed, as found in the LICENSE file.