UGV
Fengtai Wanda Plaza Indoor Localization
Large-scale indoor visual localization
Three-floor shopping-mall dataset with 679 RGB-D panoramas, 2,664 smartphone queries, depth images, an aligned 1.2-billion-point reconstruction, and precise query-camera poses.
Access links
Primary places to inspect, download, read about, or reproduce the dataset.
Quick facts
- Vehicle type
- UGV
- Environment
- Indoor
- Data origin
- Real-world
- Sensor count
- 5
- Ground truth
- Available
- 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.
| Scope | Method | Reference system | Count | Rate | Coverage |
Reference provenance
|
|---|---|---|---|---|---|---|
| 6DoF | LiDAR SLAM | NavVis M6 trajectory constrained by surveyed total-station anchor points | 1 total station | Not reported | Full |
Reference provenance
Hybrid
Accuracy and notes
|
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
- Checkerboard and total-station-localized camera positions
Calibration data
Calibration reproducibility details are not reported.
Notes
Each phone was intrinsically calibrated with a checkerboard and extrinsically calibrated using total-station-assisted EPnP, but the public documentation does not establish that calibration parameter files are released.
Known or intentional limitations
Reported constraints and characteristics to check before using the dataset.
Some shop interiors are absent from the 3D model because they were not entered during scanning.
Repetitive structures and differences in devices, viewpoints, routes, and capture times make localization challenging.
The current release page and paper disagree on panorama-depth resolution, documenting 8192x4096 and 1024x512 respectively.
Cameras
Camera 1
- Model
- Not reported
- Setup
- Rgbd
- Count
- 1
- Effective cameras
- 2
- Modality
- Rgbd
- Resolution
- 8192 x 4096
- Rate
- Not reported
- Shutter
- Not Reported
- Lens type
- Not Reported
- HFOV
- 360 deg
- VFOV
- Not reported
- Notes
- One released panoramic RGB-D database product. Each of the 679 panoramas also yields 36 perspective RGB-D images with a 60-degree field of view. The current release page documents 8192x4096 panorama depth files, while the paper reports 1024x512 depth panoramas.
Camera 2
- Model
- Huawei Honor10, Mate8, and Mate9
- Setup
- Mono
- Count
- 3
- Effective cameras
- 3
- Modality
- Rgb
- Resolution
- Not reported
- Rate
- Not reported
- Shutter
- Not Reported
- Lens type
- Not Reported
- HFOV
- Not reported
- VFOV
- Not reported
- Notes
- The 2,664 query images were captured at about 1.6 m along routes and at times different from the database capture.
IMUs
Not reported
GNSS
Not reported
LiDAR
Not reported
Additional sensors
Not reported
Annotations
Reported annotation availability and task support.
- Availability
- Available
- Format
- 16-bit PNG depth images in millimetres, plus an aligned PLY point cloud with moving objects removed
- Class count
- Not reported
Semantic segmentation
Not available
Instance segmentation
Not reported
Object detection
Not reported
Optical flow
Not reported
Depth ground truth
Available
Citation
Citation key and BibTeX kept at the end of the page for reference.
@article{Liu2022Fengtai, author={Liu, Yuchen and Gao, Wei and Hu, Zhanyi}, title={A Large-Scale Dataset for Indoor Visual Localization With High-Precision Ground Truth}, journal={The International Journal of Robotics Research}, volume={41}, number={2}, pages={129--135}, year={2022}, doi={10.1177/02783649211052064}}Dataset corrections
Notice a specific issue after reviewing the page?
Report an inaccuracy