Deep Radar Detector

TL;DR

This paper introduces a deep learning approach for radar data processing that operates directly on complex radar signals, improving detection performance and enabling real-time object classification without extensive calibration.

Contribution

It presents a novel deep learning method for radar processing that uses radar calibration data for training and introduces new augmentation techniques, addressing data scarcity issues.

Findings

Superior detection performance over classical methods

Maintains real-time processing capabilities

Reduces calibration overhead

Abstract

While camera and LiDAR processing have been revolutionized since the introduction of deep learning, radar processing still relies on classical tools. In this paper, we introduce a deep learning approach for radar processing, working directly with the radar complex data. To overcome the lack of radar labeled data, we rely in training only on the radar calibration data and introduce new radar augmentation techniques. We evaluate our method on the radar 4D detection task and demonstrate superior performance compared to the classical approaches while keeping real-time performance. Applying deep learning on radar data has several advantages such as eliminating the need for an expensive radar calibration process each time and enabling classification of the detected objects with almost zero-overhead.