List of Datasets

Handheld

ADVIO

Authentic smartphone visual-inertial odometry

Handheld smartphone recordings across floors and outdoor spaces include consumer sensors, platform estimates and inertial/manual-fix camera-pose reference.

Metadata: Complete Metadata source documented Released 2018

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

Supplementary

Not reported

Quick facts

Vehicle type
Handheld
Environment
Indoor, Outdoor
Data origin
Real-world
Sensor count
9
Ground truth
Available
Calibration
Parameters Only
Annotations
Not reported

Sensor overview

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

Cameras

2 reported

IMUs

1 reported

GNSS

Not reported

LiDAR

Not reported

Radar

Not reported

Magnetometers

1 reported

Barometers

1 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 Other iPhone inertial navigation plus manual mapped fixes Not reported 100 Hz 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.
Hybrid
Reference notes
Reference reuses released IMU plus external position fixes, manual refinement and calibration; 25 cm is assumed fix noise, not proven trajectory accuracy.

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

Per-session camera intrinsics/extrinsics and inertial additive/multiplicative bias parameters.

Known or intentional limitations

Reported constraints and characteristics to check before using the dataset.

Twenty-three recordings span 4.47 km and 68 minutes across six places, with nineteen indoor and four outdoor recordings. Elevators, escalators, crowds and poor texture can challenge visual-only motion estimates.

Reference pose reuses the released iPhone IMU and manually mapped position fixes. Assumed 25 cm fix noise is not a measured independent trajectory-accuracy bound.

Tango video is documented at roughly 5 fps and pressure at irregular roughly 1 Hz; main-table values differ. Consult the maintained format and timestamps rather than assuming a uniform fixed rate.

Pixel supplies ARCore poses without established raw camera/IMU inputs. Tango point-cloud timing is irregular; proprietary API pose products are estimates rather than independent reference systems.

Initial NTP alignment is refined by motion correlation, but achieved final timing error and independent orientation/position accuracy remain unknown. CoreLocation is a combined platform location API.

Cameras

Camera 1

Model
Apple iPhone 6s
Setup
Mono
Count
1
Effective cameras
1
Modality
Rgb
Resolution
1280 x 720
Rate
60 Hz
Shutter
Rolling
Lens type
Not Reported
HFOV
Not reported
VFOV
Not reported

Camera 2

Model
Google Tango tablet fisheye
Setup
Fisheye
Count
1
Effective cameras
1
Modality
Grayscale
Resolution
640 x 480
Rate
Not reported
Shutter
Not Reported
Lens type
Not Reported
HFOV
Not reported
VFOV
Not reported
Notes
Official format documentation describes approximately 5 fps at 640x480; the main table reports 60 Hz. Fixed rate remains unknown; archive preview confirms the separate Tango video.

IMUs

Apple iPhone 6s integrated IMU

Model
Not reported
Count
1
Accel / gyro
100 / 100 Hz

GNSS

Not reported

LiDAR

Not reported

Radar

Not reported

Additional sensors

Barometer

Model
Apple iPhone 6s integrated barometric altimeter
Count
1
Notes
Maintained format documentation corroborates irregular roughly 1 Hz sampling; the main paper table says approximately 10 Hz. No fixed native rate is asserted.

Odometry

Name
Apple ARKit pose
Count
1
Rate
60 Hz
Notes
Released iPhone platform-estimated pose; separate from the hybrid evaluation reference.

Odometry

Name
Google Tango poses
Count
1
Rate
60 Hz
Notes
Raw-tracking and area-learning pose variants from the same device; not two additional physical sensors.

Odometry

Name
Google ARCore pose
Count
1
Rate
30 Hz
Notes
Released Pixel pose output only; raw Pixel camera/IMU inputs are not established.

Odometry

Name
CoreLocation platform location
Count
1
Notes
Released API location estimate may combine GNSS/network sources; a separate pure-GNSS hardware stream is not established.

Other sensor

Name
Tango depth point-cloud measurements
Count
1
Notes
The released archive contains timestamped Tango point-cloud CSVs with irregular sampling. Exact depth-imager model/count and image geometry are not established.

Magnetometer

Model
Apple iPhone 6s integrated magnetometer
Count
1
Rate
100 Hz
Notes
Uncalibrated readings.

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
cortes2018advio

BibTeX

@misc{cortes2018advio,
  author = {{Santiago Cortés} and {Arno Solin} and {Esa Rahtu} and {Juho Kannala}},
  title = {ADVIO: An Authentic Dataset for Visual-Inertial Odometry},
  year = {2018},
  eprint = {1807.09828},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1807.09828v1}
}

Dataset corrections

Notice a specific issue after reviewing the page?

Report an inaccuracy