AI prototype for automated roof plan analysis
Budmat is a Polish manufacturer of roofing systems, metal profiles, and building materials for both residential and industrial use.
Client:
Budmat
Country:
Polen
Introduction
To explore how AI can streamline building design, Budmat collaborated with Euvic to develop RoofPlanAI, a prototype that automates the analysis of roof plans using computer vision. The results showed that machine learning can interpret drawings with high precision, laying the foundation for future automation in the construction sector.
Challenge
Budmat wanted to explore whether roof plan analysis, a crucial yet highly manual step in architectural and construction workflows, could be automated with AI.
Today, analyzing roof plans involves manually interpreting complex drawings to extract dimensions, angles, and slopes. This process is not only time-consuming but also prone to human error.
To address this, Budmat aimed to test whether machine learning could detect, segment, and structure roof plan data automatically. The long-term goal was to lay the groundwork for future automation that could reduce design errors and significantly accelerate project preparation.
Solution
Euvic developed a prototype solution, called RoofPlanAI, to validate the feasibility of automating roof plan interpretation using computer vision and machine learning.
The prototype combined object detection and segmentation models to analyze roof plans and extract relevant geometric data.
Workflow & approach:
- Used YOLO to detect key roof dimensions and angles.
- Applied segmentation techniques to identify and separate individual roof surfaces (slopes).
- Extracted detailed geometric data such as dimensions, slope angles, and roof sections.
The solution was built using Python, PyTorch, OpenCV, Scikit-learn, and Pandas.
This was not a full-scale product, but rather a proof of concept to test whether AI could handle this type of visual and geometric analysis. And to what extent the results could match manual accuracy.
Result
The results were highly promising. The prototype confirmed that automation of roof plan analysis is technically feasible with modern AI tools.
RoofPlanAI demonstrated the ability to accurately detect dimensions, angles, and roof slopes from digital drawings, showing that machine learning can interpret architectural plans with increasing precision.
The proof of concept indicated clear potential benefits:
- Significant time savings compared to manual analysis.
- Improved accuracy and consistency in design data.
- A solid foundation for developing a production-ready automation system.
Summary
The RoofPlanAI prototype validated Budmat’s vision of bringing AI-driven automation into the construction and design process.
For Euvic, the project served as another demonstration of the company’s ability to:
- Rapidly test and validate data and AI use cases.
- Apply computer vision and machine learning to real-world industry challenges.
- Help clients de-risk innovation by using practical proofs of concept before scaling to full solutions.
This case highlights how AI can transform traditional industries like construction. Turning manual, repetitive tasks into intelligent, automated workflows that enhance both speed and accuracy.
