3D Poses in the Wild Dataset

Recovering Accurate 3D Human Pose in The Wild Using IMUs and a Moving Camera

Timo von Marcard¹, Roberto Henschel¹, Michael Black², Bodo Rosenhahn² and Gerard Pons-Moll³

¹TNT, Leibniz University of Hannover
²Max Planck Institute for Intelligent Systems
³Max Planck Institute for Informatics, Saarland Informatics Campus

ECCV 2018 Munich, Germany

Video

Description

The "3D Poses in the Wild dataset" is the first dataset in the wild with accurate 3D poses for evaluation. While other datasets outdoors exist, they are all restricted to a small recording volume. 3DPW is the first one that includes video footage taken from a moving phone camera.

The dataset includes:

60 video sequences.
2D pose annotations.
3D poses obtained with our method. Our method leverages video and IMU and the poses are very accurate despite the complexity of the scenes.
Camera poses for every frame in the sequences.
3D body scans and 3D people models (re-poseable and re-shapeable). Each sequence contains its corresponding models.
18 3D models in different clothing variations.

Download

For further information about the 3DPW dataset and for download links, please click here

3DPW Challenge

In Conjuction with ECCV 2020, Gerard Pons-Moll, Michael Black , Angjoo Kanazawa and Aymen Mir organized a 3DPW challenge. The challenge aims to advance the state of the art in 3D human pose estimation in the wild by standardizing protocols and metrics for 3D Pose Estimation, so that researchers compare their methods in a consistent manner in future publications.

The Evaluation Protocol for the challenge makes use of the entire 3DPW dataset. Participants could not use the 3DPW dataset for training. The evaluation servers established for the challenge remain open and serve as a benchmark for 3D Pose Estimation. To submit results to the evaluation server, please head over to the challenge website

Stanford Social Motion Forecasting

The 3DPW dataset is also being used by researcher at the Stanford Vision and Learning Lab for a Social Motion Forecasting Challenge (SoMoF). The goal of the SoMoF benchmark is to predict future human trajectories and skeleton poses using information about the surrounding scene and the other humans involved. Participants are given video of the scene and labeled trajectories and poses up to some time t and must predict trajectories and poses for all individuals through some time t+T.

The Stanford Social Motion Forecasting Challenge uses a processed version of the 3DPW Dataset for evaluation, which can be downloaded from the following link. To participate in the challenge, please head over the SoMoF website

Citation

If you use this dataset, you agree to cite the corresponding ECCV'18 paper:


    @inproceedings{vonMarcard2018,
    title = {Recovering Accurate 3D Human Pose in The Wild Using IMUs and a Moving Camera},
    author = {von Marcard, Timo and Henschel, Roberto and Black, Michael and Rosenhahn, Bodo and Pons-Moll, Gerard},
    booktitle = {European Conference on Computer Vision (ECCV)},
    year = {2018},
    month = {sep}
    }