Text2CAD: Generating Sequential CAD Designs from Beginner-to-Expert Level Text Prompts

arXiv 2024 49.9 method

TLDR

Text2CAD is the first AI framework for generating parametric CAD models from text prompts, using a transformer and a new annotated dataset.

Reasoning

The paper introduces a novel text-to-CAD framework with a large annotated dataset, showing strong relevance to parametric CAD generation. However, it lacks real-world validation and does not address B-Rep or CSG explicitly.

Read-first score

Read-first score 49.9, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 85.

Recency 8%

75.1

Uses a gentle age decay so recent papers surface without erasing older foundations. 2024

Methodology quality 25%

Screens visible abstract and analysis fields for experiment, dataset, baseline, metric, and limitation evidence. markers=dataset,evaluation,metric

Topical relevance 42%

53.1

Uses existing LLM keyword relevance scores normalized to 0-100. AI for CAD,computer-aided design,neural CAD,generative CAD,parametric CAD,B-Rep,boundary representation,constructive solid geometry,CSG,sketch extrusion,CAD generation,CAD reconstruction,text-to-CAD,image-to-CAD,point cloud to CAD,CAD program

Reproducibility 25%

Screens links and visible text for paper, code, dataset, artifact, and repository signals. pdf=False; code=False; dataset=False; markers=dataset,github

Field roles

Methodology anchor

Rank sensitivity

Stability: volatile; rank range: 69.

Keyword Scores

parametric CAD

text-to-CAD

AI for CAD

generative CAD

CAD generation

computer-aided design

neural CAD

CAD program

sketch extrusion

B-Rep

boundary representation

CAD reconstruction

constructive solid geometry

CSG

image-to-CAD

point cloud to CAD

Deep Analysis

Innovations

First AI framework for text-to-parametric CAD model generation from designer-friendly instructions for all skill levels.
Data annotation pipeline using Mistral and LLaVA-NeXT to generate text prompts for the DeepCAD dataset, producing ~660K text annotations.
End-to-end transformer-based auto-regressive network for generating parametric CAD models from text.

Methodology

The framework uses a data annotation pipeline with Mistral and LLaVA-NeXT to create text prompts from the DeepCAD dataset. An end-to-end transformer-based auto-regressive network is trained to generate parametric CAD models from input text. Evaluation is performed using metrics for visual quality, parametric precision, and geometrical accuracy.

Key Results

The proposed framework shows great potential in AI-aided design applications, as evaluated through visual quality, parametric precision, and geometrical accuracy metrics.