List of Datasets

UGV

TUM-LSI

ML camera-based pose estimation

Large-scale indoor image-localization dataset captured by a six-camera mobile mapping trolley with geo-referenced 6DoF image poses.

Metadata: Complete Metadata source documented Released 2017

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

Quick facts

Vehicle type
UGV
Environment
Indoor
Data origin
Real-world
Sensor count
6
Ground truth
Available
Calibration
Not reported
Annotations
Not reported

Sensor overview

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

Cameras

6 reported

IMUs

Not reported

GNSS

Not reported

LiDAR

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.
6DoF LiDAR SLAM Not reported Not reported Not reported Full 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 and notes
Accuracy notes
The NavVis mapping trajectory reports sub-centimeter accuracy.
Reference notes
NavVis M3 uses three Hokuyo laser range finders and SLAM to recover the mapping trajectory. The released data contains JPEG images and geo-referenced 6DoF image poses, not raw LiDAR streams.

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
Not reported

Calibration data

Calibration reproducibility details are not reported.

Notes

Not reported

Known or intentional limitations

Reported constraints and characteristics to check before using the dataset.

TUM-LSI covers one 5,575 square-meter building floor and is designed around repetitive structures and large weakly textured areas.

The complete scan contains 1,314 images from six cameras; the paper's 1,095-image evaluation subset uses only the five horizontal cameras and omits the upward-facing camera.

Images are discrete captures spaced roughly one meter apart rather than a continuous video sequence.

Dataset access requires accepting the NavVis Indoor Dataset end-user license agreement and submitting an access form.

Cameras

navvis-camera-array

Model
Panasonic 16-megapixel system camera
Setup
Mono
Count
6
Effective cameras
6
Modality
Rgb
Resolution
4592 x 3448
Rate
Not reported
Shutter
Not Reported
Lens type
Wide Angle
HFOV
Not reported
VFOV
Not reported

IMUs

Not reported

GNSS

Not reported

LiDAR

Not reported

Additional sensors

Not reported

Annotations

Reported annotation availability and task support.

Availability
Not reported
Format
Not reported
Class count
Not reported
Semantic segmentation Not reported
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
Walch2017

BibTeX

@inproceedings{Walch2017,
  title = {Image-Based Localization Using {{LSTMs}} for Structured Feature Correlation},
  booktitle = {IEEE International Conference on Computer Vision (ICCV)},
  author = {Walch, Florian and Hazirbas, Caner and Leal-Taixé, Laura and Sattler, Torsten and Hilsenbeck, Sebastian and Cremers, Daniel},
  year = {2017},
  month = oct,
  pages = {627--637},
  url = {https://arxiv.org/abs/1611.07890}
}

Dataset corrections

Notice a specific issue after reviewing the page?

Report an inaccuracy