How To Get RADDET Dataset

In this tutorial you will learn how to download and convert the RADDet dataset that can be processed by Intel® Metro AI Suite Sensor Fusion for Traffic Management.

RADDet Dataset

Download

Download dataset from website: https://github.com/ZhangAoCanada/RADDet

Refer to how to use in the website above to download the dataset.

Then from BaiduPan to download the corresponding ADC data.

Download the dataset and arrange it as the following directory tree:

|-- raddet_adc
	|-- ADC
		|-- ******.npy
		|-- ******.npy
|-- sensors_para
	|-- registration_matrix
		|-- registration_matrix.npy
	|-- stereo_para
		|-- left_maps.npy
		|-- ProjLeft.npy
		|-- ProjRight.npy
		|-- Q.npy
		|-- right_maps.npy
		|-- roiL.npy
		|-- roiR.npy
|-- test
	|-- RAD
		|-- part1
			|-- ******.npy
			|-- ******.npy
		|-- part2
			|-- ******.npy
			|-- ******.npy
	|-- gt
		|-- part1
			|-- ******.pickle
			|-- ******.pickle
		|-- part2
			|-- ******.pickle
			|-- ******.pickle
	|-- stereo_image
		|-- part1
			|-- ******.jpg
			|-- ******.jpg
		|-- part2
			|-- ******.jpg
			|-- ******.jpg
|-- train
	|-- RAD
		|-- part1
			|-- ******.npy
			|-- ******.npy
		|-- part2
			|-- ******.npy
			|-- ******.npy
	|-- gt
		|-- part1
			|-- ******.pickle
			|-- ******.pickle
		|-- part2
			|-- ******.pickle
			|-- ******.pickle
	|-- stereo_image
		|-- part1
			|-- ******.jpg
			|-- ******.jpg
		|-- part2
			|-- ******.jpg
			|-- ******.jpg

Dataset description

The details of the data capture is shown below.

"designed_frequency":			76.8 Hz,
"config_frequency":			77 Hz,
"range_size":				256,
"maximum_range":			50 m,
"doppler_size":				64,
"azimuth_size":				256,
"range_resolution":			0.1953125 m/bin,
"angular_resolution":			0.006135923 radian/bin,
"velocity_resolution":			0.41968030701528203 (m/s)/bin

The dataset has totally 6 categories, different input formats and ground truth formats. All the information that stored in the dataset can be concluded as follow.

RAD:		3D-FFT radar data with size (256, 256, 64)
stereo_image:	2 rectified stereo images
gt:		ground truth with {"classes", "boxes", "cart_boxes"}
sensors_para: 	"stereo_para" for stereo depth estimation, and "registration_matrix" for cross-sensor registration

Note: for the classes, they are ["person", "bicycle", "car", "motorcycle", "bus", "truck" ].

Also Note: for the boxes, the format is [x_center, y_center, z_center, w, h, d].

Also Note: for the cart_box, the format is [x_center, y_center, w, h].

Also Note: for the stereo_para, left_maps.npy and right_maps.npy are derived from cv2.initUndistortRectifyMap(...); all other matrices are derived from cv2.stereoRectify(...).

License: MIT License

Generate left image and radar data, then convert to bin file as multi sensor input

cd $PROJ_DIR/deployments/raddet_tools

# If you already source python virtual environment during `edgesoftware install`, you can skip this
python3 -m venv raddet-test
source raddet-test/bin/activate

python3 -m pip install --upgrade pip
pip install -r requirements.txt

export RADDET_DATASET_ROOT=/path/to/datasets/RADDet
# dataset org:

cd scripts
bash process_dataset.sh $RADDET_DATASET_ROOT

Upon success, the radar data and the corresponding left image will be extracted, the directory tree should be as follows:

|-- bin_files_v1.0
	|-- bgr
		|-- ******.bin
		|-- ******.bin
	|-- radar
		|-- ******.bin
		|-- ******.bin
|-- raddet_adc
	|-- ADC
		|-- ******.npy
		|-- ******.npy
|-- sensors_para
	|-- registration_matrix
		|-- registration_matrix.npy
		|-- registration_matrix.bin
	|-- stereo_para
		|-- left_maps.npy
		|-- left_maps.bin
		|-- ProjLeft.npy
		|-- ProjLeft.bin
		|-- ProjRight.npy
		|-- ProjRigh.bin
		|-- Q.npy
		|-- Q.bin
		|-- right_maps.npy
		|-- right_maps.bin
		|-- roiL.npy
		|-- roiL.bin
		|-- roiR.npy
		|-- roiR.bin
|-- test
	|-- RAD
		|-- part1
			|-- ******.npy
			|-- ******.npy
		|-- part2
			|-- ******.npy
			|-- ******.npy
	|-- gt
		|-- part1
			|-- ******.pickle
			|-- ******.pickle
		|-- part2
			|-- ******.pickle
			|-- ******.pickle
	|-- stereo_image
		|-- part1
			|-- ******.jpg
			|-- ******.jpg
		|-- part2
			|-- ******.jpg
			|-- ******.jpg
	|-- left
		|-- part1
			|-- ******.jpg
			|-- ******.jpg
		|-- part2
			|-- ******.jpg
			|-- ******.jpg
|-- train
	|-- RAD
		|-- part1
			|-- ******.npy
			|-- ******.npy
		|-- part2
			|-- ******.npy
			|-- ******.npy
	|-- gt
		|-- part1
			|-- ******.pickle
			|-- ******.pickle
		|-- part2
			|-- ******.pickle
			|-- ******.pickle
	|-- stereo_image
		|-- part1
			|-- ******.jpg
			|-- ******.jpg
		|-- part2
			|-- ******.jpg
			|-- ******.jpg
	|-- left
		|-- part1
			|-- ******.jpg
			|-- ******.jpg
		|-- part2
			|-- ******.jpg
			|-- ******.jpg

Where bin_files_v1.0 stores all the bin files as multi sensor input.