01
6D Pose dataset
Create a 6D Pose Annotation Dataset in TjMakeBot
Use this guide to prepare labels and model assets, add 6D object poses, validate projection quality, and export JSON/BOP files for downstream training or evaluation.
8 min6D Posemodel assetsBOP export
Prepare labels, camera data, and model assets
6D Pose data is only useful when labels, camera intrinsics, object scale, and model assets are consistent before annotation begins.
02
Define real object labels and keep object IDs stable per image.
03
Prepare GLB/GLTF model assets and default object scale for each class.
04
Collect camera intrinsics when available; otherwise review projection quality carefully.
Add and edit 6D object poses
The 6D Pose panel is designed for practical editing: choose the class, assign the model asset, adjust pose values, and use the projection overlay as feedback.
01
Open the 2D editor and switch the toolbar to 6D Pose.
02
Apply the pose configuration or upload model assets from the panel.
03
Click Add object, then set object ID, label, model asset key, translation, rotation, scale, and camera intrinsics.
04
Use the projected cuboid and the 3D preview to confirm the object position. Drag the projection when the visible placement needs correction.
Validate quality and export JSON/BOP
Do not export a 6D Pose dataset until projection, model assets, object IDs, and camera parameters pass the quality panel.
01
Resolve errors for missing model assets, invalid depth, invalid intrinsics, duplicate object IDs, and invisible projections.
02
Refresh or reopen the project to confirm saved poses load back correctly.
03
Export the dataset and inspect annotations_pose6d.json, quality_report.json, and the BOP scene files before delivery.
FAQ
Can I train 6D Pose models directly in this MVP? Not yet. Export the 6D Pose JSON/BOP package and use it in your downstream training or evaluation pipeline.
Do I need model assets for every object? For a reliable dataset, yes. Model assets make projection review, scale checks, and downstream BOP-style export much more useful.
Continue with these guides
Suggested next learning steps
If you are new, start with annotation and export guides.
If you are preparing robotics or embodied AI data, read the 3D point cloud and 6D Pose guides next.
If you are preparing delivery-grade workflows, move into review, QA, dataset versions, collaboration, pricing, and OpenClaw workflow guides.
Next step
Move from content into product action
If this guide already solved the current question, use the entry points below to continue with the actual task.