Open3D Semantic KITTI visualization tool

• This tool is modified from [https://github.com/windowsub0406/KITTI_Tutorial], and use Semantic KITTI dataset from [http://semantic-kitti.org]
• The class difinations are in their ICCV 2019 paper "SemanticKITTI: A Dataset for Semantic Scene Understanding of LiDAR Sequences" [https://arxiv.org/abs/1904.01416]
• The tool supports voxel based point cloud downsampling, FOV and range based filtering.

Installation

Download Semantic KITTI label file from [http://semantic-kitti.org/assets/data_odometry_labels.zip] and unzip it along with KITTI odometry dataset [http://www.cvlibs.net/datasets/kitti/eval_odometry.php].

conda create --name pcd python=3.6
conda activate pcd
conda install -c open3d-admin open3d=0.9.0.0
pip install opencv-python==4.2.0.32 tqdm numpy pillow

Run

export KITTI_ROOT=PATH_TO_KITTI/odometry/dataset/
python vis_velo.py --cfg config/ego_view.json --voxel 0.1
python vis_velo.py --cfg config/top_view.json --voxel 0.1

Create a new view from existing ones and modify the viewing angle.

export KITTI_ROOT=PATH_TO_KITTI/odometry/dataset/
cp config/ego_view.json config/new_view.json
# Press [Q] to save the view
# Then you can modify the FOV and range filter arguments in new_view.json
python vis_velo.py --cfg config/new_view.json --voxel 0.1 --modify