Automotive
MUSES
Semantic and uncertainty-aware panoptic segmentation
A Swiss driving dataset with 2500 annotated multimodal samples combining RGB images, events, MEMS lidar, scanning radar and IMU/GNSS data across clear weather, fog, rain, snow, day and night.
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
- 5
- Ground truth
- Not reported
- Calibration
- Not reported
- 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
- Not reported
- Processed parameters
- Not reported
- Raw calibration data
- Not reported
- Calibration targets
- Metrology-grade checkerboard
Calibration data
Calibration reproducibility details are not reported.
Notes
Vendor lidar/radar intrinsics; frame/event camera intrinsics calibrated with checkerboard and OpenCV. Pairwise camera/event/lidar extrinsics are optimized in a triangular pose graph. Radar-lidar rotation uses scan matching and translation is measured; IMU/GNSS calibration uses u-center and point-cloud consistency optimization. The paper describes calibration methods without clearly specifying released raw calibration material.
Known or intentional limitations
Reported constraints and characteristics to check before using the dataset.
Test labels are withheld; evaluation uses a submission system.
Driving scenes are exclusively Swiss, with possible Western European geographic bias.
Fog can severely reduce lidar returns; some annotated objects have no lidar points.
Faces and license plates are blurred.
Difficulty labels describe ambiguity under the RGB-only annotation stage; their aleatoric interpretation is qualified for multimodal models.
Cameras
frame
- Model
- TRI023S-CC
- Setup
- Mono
- Count
- 1
- Effective cameras
- 1
- Modality
- Rgb
- Resolution
- 1920 x 1080
- Rate
- 30 Hz
- Shutter
- Global
- Lens type
- Not Reported
- HFOV
- 77 deg
- VFOV
- 43 deg
- Notes
- Undistorted 8-bit PNGs with automatic exposure; identifiable faces and license plates are blurred. Table 2 reports the recording frequency, not the cadence of the 2500 selected annotated samples.
event
- Model
- Prophesee GEN4.1
- Setup
- Mono
- Count
- 1
- Effective cameras
- 1
- Modality
- Event
- Resolution
- 1280 x 720
- Rate
- Not reported
- Shutter
- Not Reported
- Lens type
- Not Reported
- HFOV
- 64 deg
- VFOV
- 39 deg
- Notes
- Undistorted event tuples in compressed HDF5, delivered as the three seconds preceding frame-camera exposure end. Contrast sensitivity is adjusted by illumination and event rate is controlled. The table's 15M events/s specification is not a frame rate.
IMUs
Not reported
GNSS
simpleRTK2B Fusion
- Model or name
- simpleRTK2B Fusion
- Count
- 1
- Rate
- 30 Hz
- GNSS type
- GPS_IMU
- Position output
- Available
- Velocity output
- Available
- Heading output
- Available
LiDAR
RS-LiDAR-M1
- Model or name
- RS-LiDAR-M1
- Count
- 1
- Dimensions
- 3
- Rate
- 10 Hz
- Channels
- Not reported
- Range
- 200 m
Radar
Navtech CIR-DEV
- Count
- 1
- Released data
- Scanning power images
- Frame rate
- 4 Hz
- Doppler velocity
- Not reported
- Elevation angle
- Not reported
- Configured range
- 330 m
- Field of view
- Not reported
- File format
- Range-azimuth PNG
- Notes
- FMCW scanning radar with a cosec dish; PNG data contain per-azimuth timestamp, sweep counter and received power in dB. Table 2 reports 43.8 mm range resolution and 0.9-degree horizontal angular resolution. The experiment's 150 m projection cutoff and assumed ground height do not establish a release range cutoff or measured elevation.
Pitot tube
Not reported
Additional sensors
Not reported
Annotations
Reported annotation availability and task support.
- Availability
- Available
- Format
- Not reported
- Class count
- 19
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.
Not reported
Dataset corrections
Notice a specific issue after reviewing the page?
Report an inaccuracy