crosstyan
ed721729fd
Add end-to-end RGB-D reconstruction support across the C++ core and Python API. - add a native merge_rgbd_views path, view-aware 3D pose containers, and nanobind bindings - expose Python helpers to sample aligned depth, apply per-joint offsets, lift UVD poses to world space, and run reconstruct_rgbd - add RGB-D regression tests for merging, manual pipeline parity, symmetric depth sampling windows, and out-of-bounds joints - bump the project version from 0.1.0 to 0.2.0 for the new feature surface
RapidPoseTriangulation
Fast triangulation of multiple persons from multiple camera views.
A general overview can be found in the paper RapidPoseTriangulation: Multi-view Multi-person Whole-body Human Pose Triangulation in a Millisecond.
Build
-
Clone this project:
git clone https://gitlab.com/Percipiote/RapidPoseTriangulation.git cd RapidPoseTriangulation/ -
Enable GPU-access for docker building:
-
Install nvidia container tools: Link
-
Run
sudo nano /etc/docker/daemon.jsonand add:{ "runtimes": { "nvidia": { "args": [], "path": "nvidia-container-runtime" } }, "default-runtime": "nvidia" } -
Restart docker:
sudo systemctl restart docker
-
-
Build docker container:
docker build --progress=plain -t rapidposetriangulation . ./run_container.sh -
Build triangulator:
cd /RapidPoseTriangulation/ uv sync --group dev uv run pytest tests/test_interface.py uv build
Citation
Please cite RapidPoseTriangulation if you found it helpful for your research or business.
@article{
rapidtriang,
title={{RapidPoseTriangulation: Multi-view Multi-person Whole-body Human Pose Triangulation in a Millisecond}},
author={Bermuth, Daniel and Poeppel, Alexander and Reif, Wolfgang},
journal={arXiv preprint arXiv:2503.21692},
year={2025}
}