| --- |
| pretty_name: FoundationPose Paired Synthetic Renders |
| configs: |
| - config_name: gso-scenes |
| data_files: |
| - split: train |
| path: gso/scenes/train-*.parquet |
| - split: validation |
| path: gso/scenes/validation-*.parquet |
| - config_name: gso-assets |
| data_files: |
| - split: train |
| path: gso/assets/train-*.parquet |
| - config_name: objaverse-scenes |
| data_files: |
| - split: train |
| path: objaverse/scenes/train-*.parquet |
| - split: validation |
| path: objaverse/scenes/validation-*.parquet |
| - config_name: objaverse-assets |
| data_files: |
| - split: train |
| path: objaverse/assets/train-*.parquet |
| --- |
| |
| # FoundationPose paired synthetic renders |
|
|
| Each scene row contains two synchronized views and per-object pose, mask ID, |
| bounding box, visibility, and occlusion annotations. Depth is stored as the |
| original float32 NPY bytes; RGB and uint32 instance masks are stored as PNG |
| bytes. Matrices use row-major flattened arrays and canonical column-vector |
| names (`camera_from_world` and `world_from_object`). Raw source transforms are |
| retained alongside them. |
|
|
| The assets tables contain stable source identifiers only. The render archives |
| reference external USD geometry and do not include reusable mesh assets. |
|
|
