178 objects across 7 scenes provide paired posed multi-view images, SLAM point clouds, and complete 3D shape annotations. The dataset from Facebook captures real-world challenges like occlusions and clutter. It was last updated on January 21, 2026.
Use Cases
- Benchmark 3D reconstruction algorithms based on paired multi-view images and ground-truth shapes
- Evaluate SLAM system performance based on provided point clouds
- Train models for object segmentation in cluttered scenes based on annotations
- Study occlusion handling in computer vision based on real-world scene data
Strengths
- 178 individually annotated 3D objects provide a substantial evaluation set
- 7 diverse real-world scenes introduce varied environmental conditions
- Paired data types (images, point clouds, shapes) enable multi-modal evaluation
Limitations
- Column-level documentation is absent; field semantics must be inferred after download
- Row count is unknown, which may limit suitability assessment
Provenance
- Source
- Facebook
- Collection Method
- Captured in-the-wild sequences with annotations.
- Freshness
- 2026-01-21 03:24:04