Recap-OmniCAD: A Large-Scale CAD Dataset for B-Rep and Command Sequences

Recap-OmniCAD is a large-scale computer-aided design dataset introduced in a CVPR 2026 paper. The dataset, created by authors including Dacheng Qi and Chenyu Wang, appears to unify B-Rep and command sequence representations for CAD models. Its release is announced as coming soon, with the last metadata update on March 12, 2026.

Use Cases

• Training machine learning models for CAD generation based on the described unification of B-Rep and command sequences.
• Benchmarking CAD reconstruction algorithms using the large-scale dataset mentioned in the description.
• Researching pointer-based selection mechanisms for CAD edges and faces as suggested by the associated paper title.

Strengths

• Associated with a peer-reviewed CVPR 2026 paper, indicating academic rigor.
• Described as a 'large-scale' CAD dataset, suggesting substantial size.

Limitations

• Dataset release is 'coming soon', so the actual data is not yet available for inspection or use.
• Specific details like column definitions, row count, file formats, and license are unknown.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

Introduced in the CVPR 2026 paper 'Pointer-CAD: Unifying B-Rep and Command Sequences via Pointer-based Edges & Faces Selection'.

Freshness

Last updated 2026-03-12 04:20:17; freshness should be verified.