Adaptive automotive radar data acqusition using object detection
Please check the paper for detailed understanding of the algorithm: https://arxiv.org/pdf/2010.02367.pdf
This code runs object detection on images and stores classes and bounding boxes as .pickle file. To run object detection code, activate virtual environment: source detectron2-venv/bin/activate "python3 object_detection.py --scene=1 --direction=front" In the current setup, we have 3 scenes with images from front and rear.
For coordinate transform from image to radar, "python3 coordinate_transform.py --scene=1 --direction=front"
The src_cs folder consists of the matlab code to run compressed sensing Run the following command for baseline uniform sampling reconstruction using CS
"matlab -nodesktop -nosplash -r "meas='BPD';run compressive_sensing_radar_v1(2,meas)""
We are currently working on making the code public. Check back in later for the code to run A-1 and A-2 from our paper.