Open Source RoboticsRobots, software & learning

Industrial Plenoptic Dataset (IPD)

Software

Industrial-object pose-estimation dataset and evaluation toolkit, with an accessible BOP edition and conflicting license declarations. [1]

Installation & first result →Official website

For
Researchers evaluating industrial robot perception [1]
Source rights
Named licenses are documented by part. Includes noncommercial terms. Review licenses
Unknowns
operating system, compute requirements not fully established. Review gaps

Source researched · Not tested by us
Sources checked · Research details

Licenses & related hardware

Requirements · Research details and dates

On this page

Overview

IPD supports evaluation of industrial-part pose estimation with calibrated images, object models and pose annotations. Its original repository also provides readers, rendering and robot-consistency evaluation tooling. The practical starting point now requires choosing an edition: the original download bucket failed our checks, while the BOP benchmark hosts an accessible version with a different object and camera scope. [1] [4] [8] [11]

Is this for you?

Known limitations
  • Original download URLs checked in this review return NoSuchBucket.
  • BOP and original dataset editions differ in object and camera scope.
  • License declarations conflict within the original repository and across BOP/Hugging Face.
  • Only a bounded BOP sample was inspected; no pose-estimation evaluation was run.

Editorial assessment from cited documentation; no runtime testing.

[1] [8] [9] [10] [11] · Editorial assessment

What you need

Languages / runtimes
  • Python

[2]

Dependencies
  • For the original toolkit: Python >=3.8 and dependencies in pyproject.toml
  • For the BOP edition: archive tooling and a BOP-compatible reader

Editorial assessment from cited documentation; no runtime testing.

[2] [8] · Editorial assessment

Installation & first result

  • Choose the original toolkit or BOP dataset edition explicitly.
  • Review the conflicting artifact-license declarations.
  • Download BOP base/models and inspect camera parameters and object IDs.
  • Use the validation images and matching annotations for the selected camera.

Editorial assessment from cited documentation; no runtime testing.

[8] [9] [10] · Editorial assessment

Installation

Inspect the BOP base archive and models first; acquire the validation archive separately if its size and terms fit the intended study.

Editorial assessment from cited documentation; no runtime testing.

[8] · Editorial assessment

Hardware, models & data

Inputs and outputs

Calibrated camera images, object meshes and ground-truth poses support evaluation of predicted industrial-part poses.

[1] [4] [8]

Project notes

Use the BOP files as their own edition

We reviewed the original toolkit at commit 75dfe72e2f194f2a0e8a82bd8268fbec09205567 and the BOP dataset at revision 1fe57dbe4807598088c145a15c5bf94084e34191. The original README describes 22 industrial parts and 13 cameras; the BOP card describes ten objects and three camera views. Do not treat the BOP files as a complete mirror of the original collection or combine those counts into one configuration. [1] [6] [8]

For an initial inspection, download ipd_base.zip and ipd_models.zip from the BOP dataset page. The base archive supplies camera parameters and evaluation targets; the model archive contains object meshes and model metadata. The validation archive is approximately 6.1 GB. Its image directories and annotations are camera-specific, so pair a view such as rgb_cam1 with scene_camera_cam1.json and scene_gt_cam1.json. [8] [10]

We acquired the base and model archives and selectively retrieved one validation image with its scene camera and ground-truth JSON files. The image visibly contains industrial parts on a basket fixture; the annotations expose camera intrinsics, transforms, object IDs and object poses. This confirms a meaningful accessible sample, not full dataset integrity or annotation accuracy. No training or evaluation was run. [10]

The original repository's tested Basler-LR and Photoneo archive URLs and one CAD URL returned HTTP 404 with NoSuchBucket. That is a failure of those documented links, not evidence that the dataset has disappeared everywhere. Its Python reader and download tooling remain inspectable, but the old download route should not be offered as a verified working setup. [3] [4] [11]

What we checked

Source researched only; no installation, training, benchmark or physical hardware test performed. No affiliate links or reused media.

Not yet established

  • Requirements: operating system, host computer
  • Version-specific hardware compatibility (ecosystem tags alone do not establish it)

Updates

Upstream last updated
Dec 13, 2024 [6]
Latest release
Not established
Profile updated
Oct 18, 2026
Sources last checked
Oct 18, 2026
Next review
Nov 18, 2026
Reviewed revision
commit snapshot; no stable GitHub release / 75dfe72e2f194f2a0e8a82bd8268fbec09205567 · software repository

Source files and licenses

Include as a dataset and evaluation-toolkit reference: actual BOP sample inspected, original/baseline editions distinguished, broken original links and unresolved cross-publisher license declarations explicit.

LayerAvailability / license
softwareavailable · Conflicting: README/package CC-BY-NC-SA-4.0; LICENSE.md CC-BY-NC-ND-4.0
Internal ShareAlike versus NoDerivatives conflict unresolved; do not infer commercial or modification permission. [1] [2] [7]
datasetsavailable · Conflicting: original README/package CC-BY-NC-SA-4.0; LICENSE.md CC-BY-NC-ND-4.0; BOP CC-BY-SA-4.0; HF CC-BY-4.0
Accessible BOP edition sample verified; original download bucket unavailable in scoped checks. Rights conflict unresolved; no redistribution of sample assets. [7] [8] [9] [10]
documentationavailable · Conflicting: README/package CC-BY-NC-SA-4.0; LICENSE.md CC-BY-NC-ND-4.0
Internal ShareAlike versus NoDerivatives conflict unresolved; do not infer commercial or modification permission. [1] [2] [7]

Dataset scope does not require a trained model.

Keep the rights conflict visible

The original README and package metadata declare CC-BY-NC-SA-4.0, but LICENSE.md contains CC-BY-NC-ND-4.0 terms. That is an internal ShareAlike/NoDerivatives conflict before comparing other publishers. The BOP website lists CC-BY-SA-4.0 for IPD, while the Hugging Face card declares CC-BY-4.0. This review does not resolve which declaration governs each redistributed artifact. Verify the applicable terms with the publisher before reuse that depends on those differences; catalog inclusion is not commercial-use clearance. [2] [7] [8] [9]

References

Each source links to the exact revision we read. “Used for” shows where this page relies on it.

Correction guidance · Discussion is not enabled · Profile format 0.7.0-draft.5

Change history