PDF-based construction drawing takeoff tool for detecting, identifying, and counting symbols from MEP and architectural drawings.
🚀 Live Demo: Open App
drawing-takeoff is an experimental computer vision project focused on automating quantity takeoffs from construction drawings.
The initial goal is intentionally narrow and practical:
The project is designed as a lightweight and modular foundation for future AI-assisted construction document analysis.
Manual quantity takeoffs are repetitive and time-consuming.
Construction drawings already contain:
This project explores whether computer vision and lightweight AI workflows can automate portions of the estimating and takeoff process without requiring full BIM access.
The focus is:
The first MVP (Minimum Viable Product) intentionally avoids complexity.
The goal is to validate:
“Can construction symbols be reliably counted from PDFs with lightweight CV methods?”
The initial implementation uses traditional computer vision techniques before introducing machine learning.
```text PDF → Image Rendering → Legend Extraction → Symbol Selection → Template Matching → Count Detection → CSV Export