Skip to content

Latest commit

 

History

History
12 lines (8 loc) · 627 Bytes

radar_target_detection_tsinghua.md

File metadata and controls

12 lines (8 loc) · 627 Bytes

July 2019

tl;dr: Detection on RD map. Use CNN to replace CFAR (adaptive thresholding) and binary integration (time sequence processing).

Overall impression

The paper showed that there are minimum gain of CNN over CFAR under different noise levels. And it is better than binary integration for time sequence processing.

Technical details

  • 61 frames, 0.5 ms frame time, 2 us per chirp.
  • Seems like only one object in the frame.