This repository can transform the astyx HiRes2019 Dataset into Kitti format, This is not an official tool.
Implemented and tested on Ubuntu 16.04 with Python 3.6. Note that! Only transform the groundtruth_obj3d files and calibration files.
Unzip the dataset_astyx_hires2019,the folder should look something like the following:
dataset_astyx_hires2019
dataset_astyx_demo
dataset_astyx_hires2019 #this is the root_dir
calibration
camera_front
groundtruth_obj3d
lidar_vlp16
radar_64555
dataset.json
licenes.txt
readme.txt
Add the root path of dataset in demo.py:
root_dir=' 'your_path' /dataset_astyx_hires2019/dataset_astyx_hires2019/'
run the demo.py
python demo.py
the result will find in root_dir, 'kitti_format_label' and 'kitti_format_calib'
The 'Tr_velo_to_cam' in calib file is transform matrix of radar to camera, not lidar to camera!!
"Automotive Radar Dataset for Deep Learning Based 3D Object Detection"; Meyer, Michael and Kuschk, Georg; European Radar Conference 2019