Preprocessed object meshes packaged for use with the SimVLA project. The dataset contains approximately 37 GB of data across 2,587 object directories. It is a re-distribution of meshes generated by the BODex preprocessing pipeline and was uploaded by exaFLOPs09 on Hugging Face.
Use Cases
- Training or testing robotic manipulation algorithms based on collision-friendly simplified meshes.
- Developing computer vision models for object recognition based on normalized meshes and textures.
- Building physics-based simulation environments based on URDF and USD/USDA files.
- Creating synthetic datasets for machine learning based on the collection of textured 3D models.
Strengths
- Contains 2,587 distinct object directories.
- Provides approximately 37 GB of data, indicating a substantial collection.
- Includes multiple mesh representations (simplified and normalized) and material textures per object.
Limitations
- Description metadata is limited; actual data quality requires manual inspection after download.
- Column-level documentation is absent; field semantics must be inferred after download.
- Row count is unknown, which may limit suitability assessment.
Provenance
- Source
- BODex preprocessing pipeline.
- Collection Method
- Re-distribution of generated meshes.
- Freshness
- Last updated 2026-05-21 13:40:21; freshness should be verified.