State Key Lab of CAD & CG, Zhejiang University
LightStage is a multi-view dataset, which is proposed in NeuralBody. This dataset captures multiple dynamic human videos using a multi-camera system that has 20+ synchronized cameras. The humans perform complex motions, including twirling, Taichi, arm swings, warmup, punching, and kicking. We provide the SMPL-X parameters recovered with EasyMocap, which contain the motions of body, hand and face. Please download the dataset from [here](https://github.com/zju3dv/neuralbody/blob/master/INSTALL.md#zju-mocap-dataset).
Prev
1
Next
"Mirrored-Human" is a large-scale Internet dataset containing videos in which we can see the person and the person's image through a mirror, together with pseudo ground-truth annotations generated by our optimization-based framework.
The dataset includes:
- video clips (1500k+ frames).
- 2D human keypoints.
- 3D poses (SMPL model parameters) obtained with our method. Our method leverages an additional view provided by the mirror reflection so as to resolve the depth ambiguity.
Here are some sample data:
Prev
1
Next
Please cite the corresponding paper if you use this dataset.
@inproceedings{peng2021neural,
title={Neural Body: Implicit Neural Representations with Structured Latent Codes for Novel View Synthesis of Dynamic Humans},
author={Peng, Sida and Zhang, Yuanqing and Xu, Yinghao and Wang, Qianqian and Shuai, Qing and Bao, Hujun and Zhou, Xiaowei},
booktitle={CVPR},
year={2021}
}
@inproceedings{fang2021mirrored,
title={Reconstructing 3D Human Pose by Watching Humans in the Mirror},
author={Fang, Qi and Shuai, Qing and Dong, Junting and Bao, Hujun and Zhou, Xiaowei},
booktitle={CVPR},
year={2021}
}