This repository contains the code used for the ITSC 2022 paper Extrinsic Camera Calibration with Semantic Segmentation.
A linux distribution with a working Python installation is required.
To create the Python environment, creating a Python venv is advised.
To install the required dependencies:
sudo apt install python3.7-dev ffmpeg, replace 3.7 with the Python version installed on your system.
pip3 install -r requirements.txt
To checkout the patched Cylinder3D version for automatic point cloud labeling:
git submodule update --init --recursive
In order to use the CARLA data generator, the CARLA server and Python API needs to installed, see this link for instructions.
In order to generate data from the KITTI dataset, the KITTI dataset needs to be symlinked to datasets/KITTI.
To evaluate the algorithm on CARLA data, use evaluate_carla.sh. To generate data in CARLA, uncomment the second line in evaluate_carla.sh.