Scene-aware Learning Network for Radar Object Detection

Abstract

Object detection is essential to safe autonomous or assisted driving. Previous works usually utilize RGB images or LiDAR point clouds to identify and localize multiple objects in self-driving. However, cameras tend to fail in bad driving conditions, e.g. bad weather or weak lighting, while LiDAR scanners are too expensive to get widely deployed in commercial applications. Radar has been drawing more and more attention due to its robustness and low cost. In this paper, we propose a scene-aware radar learning framework for accurate and robust object detection. First, the learning framework contains branches conditioning on the scene category of the radar sequence; with each branch optimized for a specific type of scene. Second, three different 3D autoencoder-based architectures are proposed for radar object detection and ensemble learning is performed over the different architectures to further boost the final performance. Third, we propose novel scene-aware sequence mix augmentation (SceneMix) and scene-specific post-processing to generate more robust detection results. In the ROD2021 Challenge, we achieved a final result of average precision of 75.0% and an average recall of 81.0%. Moreover, in the parking lot scene, our framework ranks first with an average precision of 97.8% and an average recall of 98.6%, which demonstrates the effectiveness of our framework. CCS CONCEPTS• Computing methodologies → Object detection; Scene understanding; Neural networks.

Towards Deep Radar Perception for Autonomous Driving: Datasets, Methods, and Challenges

“…There are some possible ways to adapt the network to different weather conditions. One way is to add a scene switching module [229], then use different networks for different weather. This method is straightforward, but introduces additional computational and memory costs.…”

DAROD: A Deep Automotive Radar Object Detector on Range-Doppler maps

“…In a similar manner, Akita et al [31] simultaneously track and classify targets using extracted ROI from RA maps. Recently, as part of the CRUW challenge [7] new object detection and classification architectures were proposed on RA maps such as [32] and [9].…”

“…Instead, raw data tensors and deep neural networks can be used to replace and improve traditional techniques for object detection, classification and segmentation without losing information. Recently, radar datasets and challenges such as CARRADA [5], RADDet [6] or CRUW [7], where radar data is provided as raw data tensors, have opened up research on new deep learning methods for automotive radar ranging from object detection [6], [8], [9] to object segmentation [10].…”