List of Datasets

UGV

RELLIS-3D

Multi-modal semantic perception for off-road navigation

Five Warthog off-road traversals with RGB, stereo, dual lidar, GNSS/INS and image/point semantic annotations.

Metadata: Complete Metadata curator reviewed Released 2020

Primary places to inspect, download, read about, or reproduce the dataset.

Quick facts

Vehicle type
UGV
Environment
Outdoor
Data origin
Real-world
Sensor count
7
Ground truth
Available
Calibration
Partial Raw Data
Annotations
Available

Sensor overview

High-level sensor availability before the detailed sensor records below.

Cameras

3 reported

IMUs

1 reported

GNSS

1 reported

LiDAR

2 reported

Radar

Not reported

Magnetometers

Not reported

Barometers

Not reported

UWB

Not reported

Ground truth

Reference scope, method, rate, coverage, and provenance.

Scope Method Reference system Count Rate Coverage Reference provenance
External
Measured using equipment independent of the platform’s normal onboard sensor pipeline.
Derived
Computed primarily from the dataset’s own recorded sensors.
Hybrid
Combines independent reference equipment with onboard measurements.
Other Manual annotation Not reported Not reported Not reported Partial · 6235 Selected RGB Frames and 13556 Lidar Scans; Raw Data Beyond Selected Labels Is Not Universally Annotated. Reference provenance
External
Measured using equipment independent of the platform’s normal onboard sensor pipeline.
Derived
Computed primarily from the dataset’s own recorded sensors.
Hybrid
Combines independent reference equipment with onboard measurements.
Derived
Accuracy notes
Human annotation verification is described; geometric trajectory error is not quantified.
Reference notes
Pixel-wise and point-wise semantic reference labels share a manually curated ontology; scan SLAM supports annotation rather than independent vehicle trajectory truth.

Calibration and synchronization

Reported calibration level, reproducibility signals, and supporting notes.

Level
Partial Raw Data
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

Official release lists camera intrinsics, Basler-to-Ouster and Velodyne-to-Ouster transforms, stereo calibration and a raw calibration package. Complete raw coverage for all sensor pairings is not established.

Known or intentional limitations

Reported constraints and characteristics to check before using the dataset.

Severe class imbalance and ambiguous off-road semantic boundaries.

Collection covers one outdoor facility.

Independent odometry ground truth is not established.

Printed split totals, evaluated class counts and KPConv result require reconciliation.

Release lists scan poses for annotation support; their presence does not establish independent odometry ground truth.

Raw calibration package is listed, but its complete sensor-pair coverage was not inspected. Download providers can impose access quotas.

Cameras

Camera 1

Model
Basler acA1920-50gc
Setup
Mono
Count
1
Effective cameras
1
Modality
Rgb
Resolution
1920 x 1200
Rate
10 Hz
Shutter
Not Reported
Lens type
Not Reported
HFOV
Not reported
VFOV
Not reported
Notes
RGB camera; 16 mm/F18 Edmund lens as printed.

Camera 2

Model
Nerian Karmin2 + SceneScan
Setup
Stereo
Count
1
Effective cameras
2
Modality
Not Reported
Resolution
Not reported
Rate
10 Hz
Shutter
Not Reported
Lens type
Not Reported
HFOV
Not reported
VFOV
Not reported
Notes
Stereo images and depth images are described; exact optical color modality/resolution and depth accuracy are not specified.

IMUs

Vectornav VN-300

Model
Vectornav VN-300
Count
1
Accel / gyro
100 / 100 Hz

GNSS

Vectornav VN-300

Model or name
Vectornav VN-300
Count
1
Rate
Not reported
GNSS type
GPS_IMU
Position output
Available
Velocity output
Not reported
Heading output
Not reported

LiDAR

Ouster OS1

Model or name
Ouster OS1
Count
1
Dimensions
3
Rate
10 Hz
Channels
64
Range
Not reported

Velodyne Ultra Puck

Model or name
Velodyne Ultra Puck
Count
1
Dimensions
3
Rate
10 Hz
Channels
32
Range
Not reported

Radar

Not reported

Additional sensors

Not reported

Annotations

Reported annotation availability and task support.

Availability
Available
Format
Image labels in color/ID formats; Ouster and Velodyne scans and labels in SemanticKITTI format, plus annotated Ouster PLY files. Native ontology includes void among twenty classes.
Class count
[void dirt grass tree pole water sky vehicle object asphalt building log person fence bush concrete barrier puddle mud rubble]
Semantic segmentation Available
Instance segmentation Not reported
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.

Citation key
jiang2022rellis3d

BibTeX

@misc{jiang2022rellis3d,
  title = {RELLIS-3D Dataset: Data, Benchmarks and Analysis},
  author = {Peng Jiang and Philip Osteen and Maggie Wigness and Srikanth Saripalli},
  year = {2022},
  eprint = {2011.12954v4},
  archivePrefix = {arXiv}
}

Dataset corrections

Notice a specific issue after reviewing the page?

Report an inaccuracy