Construction photo to drawing AI
Turn a site photo into a measured wall elevation and a materials estimate.
Live demo
A simplified, self-contained version of the core idea. Everything runs in your browser with sample data.
Calibrate with one known dimension: a standard 838 mm door leaf measured in the photo.
| Scale | 7.10 mm/px |
| Wall width | 4900 mm |
| Net board area | 8.84 m² |
| Plasterboard 1200×2400 (+10%) | 4 sheets |
| Studs @ 600 centres | 13 |
Draft — not for construction approval unless checked by a qualified professional.
The problem
Measuring walls and counting boards by hand is slow, and generative image tools produce pictures, not measurements.
Our solution
Computer-vision segmentation (YOLO) finds walls, doors and windows; one known dimension calibrates the scale; geometry is cleaned up and turned into an elevation with dimension lines. Users correct the drawing in an SVG editor, and their corrections become training data.
- Scale calibration from known dimensions
- SVG correction editor
- Exports to SVG, PDF, DXF and JSON
- Every output clearly marked as a draft for professional checking
Result
MVP covering upload, background AI processing, elevation geometry, material estimates and CAD-style exports.