Bevlshaper

bevlshaper

Algorithm for bird's-eye-view L-shape fitting in 3D LIDAR point clouds from traffic scenarios

:hammer: UNDER DEVELOPMENT :wrench:

Foreword

This project is inspired by the paper Efficient L-Shape Fitting for Vehicle Detection Using Laser Scanners. As in the paper, a K-D tree algorithm is used for segmentation after point cloud filtering. This step is inspired by Moving object classification using horizontal laser scan data. Within found clusters, L-shapes are detected. For easy prototyping and modelling, Python was used instead of a more computationally powerful language. The NumPy library is heavily used.

Prepare environment

conda create --name kitti -y python=3
conda activate kitti

Install dependencies

git clone https://github.com/scud3r1a/exploreKITTI.git
pip install -r requirements.txt

Run bevlshaper on KITTI dataset scenes

# Display traffic scene in BEV
python render_scene.py

# Run segmentation and detection algorithm
# See main.py for execution config
python main.py