Benchmark RadarOmniMech
OmniMech evaluates vision-language models on four tasks: CAD program synthesis from engineering drawings, diagram-to-3D reasoning, annotation-grounded reasoning, and tool-augmented agentic reasoning using industrial mechanical data with 251k drawings and associated CAD models.
- Released
- 2026-08-06
- Readiness
- Paper only
- Primary field
- Industrial & Engineering
Why it matters
Existing benchmarks focus on coarse 3D objects; OmniMech addresses the need for evaluating VLMs on fine-grained, dimensioned mechanical designs, providing a standardized testbed to assess progress in executable CAD generation and 3D reconstruction for industrial applications.
Motivation
Recent vision-language models (VLMs) can generate executable CAD programs from images, but existing methods mainly target coarse, general-purpose 3D objects and rarely address the fine-grained geometry and millimeter-level tolerances required in industrial mechanical design.
Primary resources
Benchmark Radar records only publicly supported details and links back to primary sources for verification.