New framework generates parametric CAD code from raster drawings

Researchers have developed Drawing-Recode, a new framework designed to generate parametric CAD code from raster 2D CAD drawings. This system addresses the limitations of existing methods by explicitly linking dimensional annotations to geometric features using cross-attention and a novel Annotation Grounding Loss. The framework employs an image encoder for feature extraction and a text recognition module for annotations, feeding the combined information into a Large Language Model to produce CAD code in the Structured Parametric CAD Code format. Experiments indicate that Drawing-Recode surpasses current baselines, even on scanned drawings representative of industrial conditions, offering a significant contribution to the digitization of legacy CAD data for manufacturing automation. AI

IMPACT This research could streamline the digitization of legacy CAD data, potentially accelerating manufacturing processes and part reproduction.

RANK_REASON Academic paper detailing a new framework for CAD code generation. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CV →

• arXiv
• computer-aided design
• Drawing-Recode
• Large Language Model

• paper
• product

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →