List of Datasets

Other

XYZ-IBD

Industrial bin-picking object detection and 6D pose estimation

Real multi-view industrial bin scenes and synthetic RGB-D training images of fifteen reflective, textureless parts, with instance masks and 6D object poses.

Metadata: Partial Metadata needs review

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

Download

Not reported

Code

Not reported

Supplementary

Not reported

Quick facts

Vehicle type
Other
Environment
Not reported
Data origin
Hybrid
Sensor count
4
Ground truth
Available
Calibration
Not reported
Annotations
Available

Sensor overview

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

Cameras

4 reported

IMUs

Not reported

GNSS

Not reported

LiDAR

Not reported

Radar

Not reported

Pitot tube

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 Manual annotation Not reported Not reported Not reported Partial · Real-World Validation 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
Simulation replication reports mean position error 0.999 mm (SD 0.12 mm) and angle error 0.432 degrees (SD 0.08 degrees); Table 2 rounds overall error to 0.99 mm. This is simulation-based pipeline validation rather than independent physical metrology of all real scenes.
Reference notes
Coated-depth fusion, manual CAD alignment and multi-scale ICP; poses propagate to 50 viewpoints and other cameras through calibrated transforms. Test poses are retained on an evaluation server.
Other Other Not reported Not reported Not reported Partial · Released Real-World Annotated Data 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
Reference notes
Enhanced depth is captured with Acksys SP-102 coating; after evaporation a native-appearance pass supplies aligned raw depth. Public test ground truth is withheld.
6DoF Other Not reported Not reported Not reported Partial · Synthetic Training 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
Reference notes
Physics-based rendered training data provide object poses, instance masks and depth maps; this product is separate from simulation used to validate real annotation error.

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
Four precisely machined spheres

Calibration data

Calibration reproducibility details are not reported.

Notes

Fifty sampled viewpoints; hand-eye initialization and sphere-cloud ICP refine viewpoint transforms. Calibration prose gives about 0.248 mm average RMSE while Table 2 gives 0.245 mm. The paper does not establish a public raw calibration release.

Known or intentional limitations

Reported constraints and characteristics to check before using the dataset.

Test images are released but test ground truth is hosted on an evaluation server.

Dense clutter, repeated instances, reflective materials and geometric symmetries are deliberate perception challenges.

Real-world annotation accuracy is supported by simulated replication of the pipeline.

Cameras

realsense-d415

Model
Intel RealSense D415
Setup
Rgbd
Count
1
Effective cameras
2
Modality
Rgbd
Resolution
1920 x 1080
Rate
Not reported
Shutter
Not Reported
Lens type
Not Reported
HFOV
Not reported
VFOV
Not reported
Notes
RGB dimensions only; depth dimensions are not established. The hardware paragraph reports 30 FPS, which does not establish cadence of the released discrete-viewpoint images.

xyz-al-m-dlp

Model
XYZ Robotic AL-M DLP
Setup
Other
Count
1
Effective cameras
1
Modality
Grayscale
Resolution
1440 x 1080
Rate
Not reported
Shutter
Not Reported
Lens type
Not Reported
HFOV
Not reported
VFOV
Not reported
Notes
Grayscale plus structured-light depth; 1440x1080 describes grayscale imagery. This is the primary camera for real pose annotation; labels are projected to the other cameras.

photoneo-phoxi-m

Model
Photoneo PhoXi M
Setup
Other
Count
1
Effective cameras
1
Modality
Grayscale
Resolution
Not reported
Rate
Not reported
Shutter
Not Reported
Lens type
Not Reported
HFOV
Not reported
VFOV
Not reported
Notes
Laser-triangulation depth and grayscale imagery. The hardware description reports up to 2064x1544 depth resolution and 0.1 mm voxel precision; these are not pose-label accuracy or delivered frame rate.

IMUs

Not reported

GNSS

Not reported

LiDAR

Not reported

Radar

Not reported

Pitot tube

Not reported

Additional sensors

Not reported

Annotations

Reported annotation availability and task support.

Availability
Available
Format
Not reported
Class count
15
Semantic segmentation Not reported
Instance segmentation Available
Object detection Available
Optical flow Not reported
Depth ground truth Available

Citation

Citation key and BibTeX kept at the end of the page for reference.

Citation key
Not reported

BibTeX

Not reported

Dataset corrections

Notice a specific issue after reviewing the page?

Report an inaccuracy