The authors of From lifestyle vlogs to everyday interactions generously shared a people-split of their VLOG dataset, where OpenPose was found to fire well, from which we used a subset to form our VLOG-people dataset. The video ids that we used can be found in datasets/vlog/vlog_ids.txt.
We provide our pre-processed tfrecords, which can be accessed here. Details of the tfrecord format can be found in doc/datasets.md.
If you use this data for your research, please cite:
@inproceedings{fouhey2018lifestyle,
title={From lifestyle vlogs to everyday interactions},
author={Fouhey, David F and Kuo, Wei-cheng and Efros, Alexei A and Malik, Jitendra},
booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition},
pages={4991--5000},
year={2018}
}
@InProceedings{humanMotionKZFM19,
title={Learning 3D Human Dynamics from Video},
author = {Angjoo Kanazawa and Jason Y. Zhang and Panna Felsen and Jitendra Malik},
booktitle={Computer Vision and Pattern Recognition (CVPR)},
year={2019}