HPS Dataset demo scripts

Python scripts for demo/visualization of various data from HPS Dataset

Requirements

• GPU with OpenGL 4.0 and EGL support
• ffmpeg>=2.1 with libx264 enabled and ffprobe installed

Installation

1. git clone https://github.com/vguzov/hps_dataset_scripts
2. pip install -r requirements.txt
3. Get the SMPL model: Follow install instructions on https://github.com/gulvarol/smplpytorch

Sample 1: visualize camera self-localization results

• Download localization results and scans from http://virtualhumans.mpi-inf.mpg.de/hps/
• Keep the scans packed
• Choose the virtual camera calibration: there are 2 choices, 029756 or 029757, which represents calibrations from real cameras with S/N 029756 and 029757. These cameras were used during the data capturing process.
• (Optional) To render split screen view, download and unpack camera videos

Run python render_visual_localization.py <path to localization json> <path to appopriate scan zip> <output mp4> --camera <choose 029756 or 029757> (to render split screen, pass -iv <path to appropriate video>)

Sample result (with split screen rendering):

Sample 2: render a moving SMPL model inside a 3D scene scan

• Go to https://github.com/vguzov/cloudrender
• Follow install instructions
• Run download_test_assets.sh and test_scene_video.py from there

In test_assets/output.mp4 you should get the video similar to this one:

Citation

If you find the code or data useful, please cite:

@inproceedings{HPS,
    title = {Human POSEitioning System (HPS): 3D Human Pose Estimation and Self-localization in Large Scenes from Body-Mounted Sensors },
    author = {Guzov, Vladimir and Mir, Aymen and Sattler, Torsten and Pons-Moll, Gerard},
    booktitle = {{IEEE} Conference on Computer Vision and Pattern Recognition (CVPR)},
    month = {jun},
    organization = {{IEEE}},
    year = {2021},
}