StrayVisualizer

A set of scripts to work with data collected using Stray Scanner.

Usage

To access the data from your device, see this wiki entry.

Installing Dependencies

Install dependencies with pip -r requirements.txt.

Example Datasets

If you don't have your own dataset, download one of these example datasets:

wget https://stray-data.nyc3.digitaloceanspaces.com/datasets/ZB1.tar.gz
wget https://stray-data.nyc3.digitaloceanspaces.com/datasets/ZB2.tar.gz
wget https://stray-data.nyc3.digitaloceanspaces.com/datasets/ZB3.tar.gz

Assuming you selected ZB1.tar.gz, you can extract the dataset using the command tar -xvf ZB1.tar.gz.

Visualizing the data

Run python stray_visualize.py <path-to-dataset>.

Available command line options are:

--point-clouds shows pointclouds.
--confidence=<value> there are three levels of confidence for the depth outputs: 0, 1 and 2. Higher is more confident. Only points having confidence higher or equal to the given value are shown.
--frames shows the camera pose coordinate frames.
--every=<n> determines how often the coordinate frame is drawn. Default is to draw every 60th frame.
--trajectory shows a black line for the trajectory of the camera.
--integrate will run the data through the Open3D RGB-D integration pipeline and visualize the resulting mesh.
--voxel-size=<size> sets the voxel size in meters for RGB-D integration.
--mesh-filename save the mesh from RGB-D integration into the given file. Defaults to no mesh saved.

Creating a Video From the Depth Maps

python make_video.py <dataset-path> will combine depth maps into a video. It will create a file depth_video.mp4 inside the dataset folder.

Running 3D Reconstruction on Collected Data

For convenience, the convert_to_open3d.py script is provided to convert the Stray Scanner format to the Open3D reconstruction system format.

Usage: python convert_to_open3d.py --dataset <path-to-dataset> --out <where-to-save-converted-dataset>.

You can run their reconstruction pipeline using python <path-to-open3d-repo>/examples/python/reconstruction_system/run_system.py <config.json> --make --register --refine --integrate as described here. <config.json> is a configuration file created for convenience by convert_to_open3d.py into the newly created dataset folder. It contains absolute paths, so if you move your dataset, be sure to update the configuration.

Beware that the Open3D reconstruction system takes up quite a lot of memory and compute. On Mac, you might need to add "python_multi_threading": false into the config file to avoid crashing.

Reporting Issues

If you find any issues with this project or bugs in the Stray Scanner app, you can open an issue on this repository.

PyojinKim / StrayVisualizer