jedeschaud / kitti_carla_simulator

KITTI-CARLA: Python scripts to generate the KITTI-CARLA dataset
MIT License
65 stars 12 forks source link
carla-simulator dataset kitti point-cloud synthetic

KITTI-CARLA: a KITTI-like dataset generated by CARLA Simulator

Python scripts to generate the KITTI-CARLA dataset (see our article)

Tested on Windows 10 with CARLA 0.9.10 in C:/CARLA_0.9.10/WindowsNoEditor

The full dataset can be downloaded here.

Installation

> git clone https://github.com/jedeschaud/kitti_carla_simulator

Running the data generation

Launch CarlaUE4
> python KITTI_data_generator.py #run the generation of a KITTI-like vehicule in the 7 basic maps of CARLA

Creating complete point clouds with the ground truth trajectory

> python KITTI_georeferencing.py #create the georefenced point cloud from LiDAR scans and ground truth trajectory

Creating complete colored point clouds

> python KITTI_colorization.py #create the colored point cloud from complete point clouds and images

Citation

If you use our work in your research project, please consider citing:

@article{deschaud2021kitticarla,
  author = {{Deschaud}, Jean-Emmanuel},
  title = "{KITTI-CARLA: a KITTI-like dataset generated by CARLA Simulator}",
  journal = {arXiv e-prints},
  keywords = {Computer Science - Computer Vision and Pattern Recognition, Computer Science - Robotics},
  year = 2021,
  eid = {arXiv:2109.00892}
}