awesome-radar-perception

A curated list of radar datasets, detection, tracking and fusion.
Keep updating.
Author: Yi Zhou
Contact: zhouyi1023@tju.edu.cn

🚩I have published a review paper on radar perception. Please see the link below. It is open access. If you find the contents are useful, please cite this paper in your work. I will keep updating this repository for the latest works in the radar perception field.

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

The 41-page slides associated with this paper: Link ; Link for China Mainland

Contents

Overview

• Review
• Seminars and Workshops

Data Perspective:

• Radar Datasets
• Radar Signature
• Calibration
• Labelling
• Augmentation
• Simulation
• Generative Model
• Testing

Signal Processing:

• Radar Toolbox
• MIMO Calibration
• Detector
• Super Resolution
• Clustering
• Denoising

Applications:

• TI Reference Designs
• Ego-Motion Estimation
• Velocity Estimation
• Depth Estimation
• Object Detection
• Sensor Fusion
• Weakly Supervised
• Tracking
• Prediction
• Occupancy Grid Map
• Open Space Segmentation
• Scene Understanding (Static Segmentation)
• Place Recognition
• Odometry and SLAM
• Automotive SAR
• Human Activity
• Radar-Audio

Challenges:

• Weather Effect
• Multi Path Effect
• Mutual Interference
• Cell Migration
• Tx-Rx Leakage
• Imperfect Waveform Separation
  

Radar Datasets

In my review paper, there is a table with more detials.

Conventional Radar Datasets for Autonomous Driving

Comments: nuScenes, DENSE and Pixset are for sensor fusion, but not particularly address the role of radar. Radar scenes provides point-wise annotations for radar point cloud, but has no other modalities. Pointillism uses 2 radars with overlapped view. Zendar seems no longer available for downloading. AiMotive focuses on long-range 360 degree multi-sensor fusion.

Pre-CFAR Datasets for Detection

Comments: CARRADA is captured in clean scenarios, CRUW uses RA maps, RADDet provides annotations for RAD tensor, RADICaL provides raw ADC data and signal processing toolboxes, GhentVRU can be accssed by contacting with authors, ODA is for drones and provides event camera data.

4D Radar Datasets

Comments: Astyx is small, VoD focuses on VRU classification, RADIal's annotation is coarse but provides raw data, TJ4D features for its long range detection, K-Radar provides RAD tensor and 3D annotations. ZF 4DRadar Dataset is not yet public available.

Specific Tasks

Odometry and Localization

Gesture

Human Activity and Reconstruction

Vital Sign

Radar Toolbox

Simulation

RadarSimPy: Code;
Virtual Radar: Code;
MaxRay: Paper

TI Signal Processing SDK:

RaDICaL's Toolbox: SDK;
PyRapid: SDK;
OpenRadar : SDK;
Pymmw: SDK;
Open radar initiative: SDK;
RADIal's Emptyband-DDM Script: Code

Official SDK:

NXP Premium Radar SDK: Link;
TI mmWAVE Studio: Link;
TI Toolbox: