Automotive
RadarScenes
Point-wise automotive radar semantics and instance tracking
Four asynchronously cycling radars, odometry and anonymized documentation images support manually labeled moving-road-user perception.
Access links
Primary places to inspect, download, read about, or reproduce the dataset.
Quick facts
- Vehicle type
- Automotive
- Environment
- Outdoor
- Data origin
- Real-world
- Sensor count
- 6
- Ground truth
- Not reported
- Calibration
- Parameters Only
- Annotations
- Available
Sensor overview
High-level sensor availability before the detailed sensor records below.
Ground truth
Reference scope, method, rate, coverage, and provenance.
Not reported
Calibration and synchronization
Reported calibration level, reproducibility signals, and supporting notes.
- Level
- Parameters Only
- Processed parameters
- Available
- Raw calibration data
- Not reported
- Calibration targets
- Not reported
Parameters only
Processed calibration parameters are reported, but raw calibration data is not reported.
Notes
sensors.json provides radar x/y mounting positions and yaw angles. Camera calibration and raw calibration sessions are not established.
Known or intentional limitations
Reported constraints and characteristics to check before using the dataset.
Eleven road-user classes plus static fallback differ from the five-class evaluation mapping. Static/background detections dominate the data.
Radars cycle independently with variable timing; approximately 60 ms is an average, not a uniform sampling interval.
Documentary images are repainted for anonymization and are intended as scene guidance. Image segmentation artifacts and unknown calibration limit camera-fusion use.
Delivered ego-motion combines vehicle odometry and DGPS, but no separate raw GNSS stream or independent navigation reference is established. Hardware models, timing error and numerical pose error remain unknown.
Cameras
Camera 1
- Model
- Not reported
- Setup
- Mono
- Count
- 1
- Effective cameras
- 1
- Modality
- Rgb
- Resolution
- Not reported
- Rate
- Not reported
- Shutter
- Not Reported
- Lens type
- Not Reported
- HFOV
- Not reported
- VFOV
- Not reported
- Notes
- Released documentary JPGs are anonymized/repainted; no camera calibration/rate assumed.
IMUs
Not reported
GNSS
Not reported
LiDAR
Not reported
Radar
Radar 1
- Count
- 4
- Released data
- Point detections / point clouds
- Frame rate
- Not reported
- Doppler velocity
- Available
- Elevation angle
- Not available
- Configured range
- 100 m
- Field of view
- 120° H (per radar)
- Carrier frequency
- 77 GHz
- File format
- HDF5
- Notes
- 77 GHz, near-range 100 m, ±60-degree view; independent varying cycles average about 60 ms (≈17 Hz).
Additional sensors
Odometry
- Name
- Vehicle motion state
- Count
- 1
- Notes
- Released x/y/yaw, forward velocity and yaw-rate table; official documentation attributes ego motion to vehicle odometry and DGPS. Exact fusion pipeline and independent accuracy remain unknown.
Annotations
Reported annotation availability and task support.
- Availability
- Available
- Format
- HDF5 point class/track IDs and UUIDs; eleven road-user classes plus static fallback, with five-class evaluation mapping
- Class count
- 11
Semantic segmentation
Available
Instance segmentation
Available
Object detection
Not reported
Optical flow
Not reported
Depth ground truth
Not reported
Citation
Citation key and BibTeX kept at the end of the page for reference.
@misc{schumann2021radarscenes,
author = {Schumann, Ole and Hahn, Markus and Scheiner, Nicolas and Weishaupt, Fabio and Tilly, Julius F. and Dickmann, Jürgen and Wöhler, Christian},
title = {RadarScenes: A Real-World Radar Point Cloud Data Set for Automotive Applications},
year = {2021},
eprint = {2104.02493},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2104.02493v2}
}Dataset corrections
Notice a specific issue after reviewing the page?
Report an inaccuracy