This is the documentation for Pre-processing of 4D Radar data with K-Radar dataset (refer to the paper).
You can download the pre-processed 4D Radar data in google drive link: URL
The density level of the pre-processed 4D Radar tensor data uploaded on Google Drive ranges from 0.01% to 10%. If you need data with different levels of density, please refer to the information below.
The first stage extracts the main measurements with high power from the 4D Radar tensor (e.g., percentile in RTNH and CFAR in common).
We provide pre-processing code in two versions: Cartesian coordinates and Polar coordinates.
1. Cartesian Coordinates
In this case, use kradar detection version 1.1. Use radar_zyx_cube as input.
Input data: 3DRT-XYZ
Output data: 4D Sparse Radar Tensor in cartesian coordinates
Modify data path and configs in configs/sparse_rdr_data_generation/cfg_gen_wider_rtnh_10p.yml
. You can modify QUANTILE_RATE
(line 34) to change the density of the data.
$ python kradar_detection_v1_1.py
2. Polar Coordinates
In this case, use kradar detection version 2.1. Use rdr_polar_3d as input.
Input data: 4DRT
Output data: 4D Sparse Radar Tensor in polar coordinates
Modify dict_cfg
in kradar_detection_v2_1.py
. You can modify rate
in def get_proportional_rdr_points
to change the density of the data.
#in __main__
1. use the preprocessed data
dict_item = kradar_detection.get_proportional_rdr_points(dict_item)
2. save the preprocessed data
kradar_detection.save_proportional_rdr_pc()
#run code
$ python kradar_detection_v2_1.py
Link Wide range, Quantile (RTNH) / 2 stage CFAR for Sidelobe filtering (RTNH+): code will be uploaded soon. (gif) Function: datasetv1_1 generate~ Density: ref Enhanced