Upload a floor plan, a scan, or a phone photo of a sketch on paper, and Strucit traces the walls, doors and windows for you, in your browser, ready to place equipment on.
How it works
- Import the plan as the reference image: a PNG, JPEG or WebP of a CAD export, a scan, a photo from your phone, or a PDF (pick the page if it has several).
- Calibrate the scale. Draw a line over a wall you know the length of, so every detected wall comes out at real size. Detection asks for this first.
- Optionally draw a detection area around the plan to skip title blocks, legends and handwritten notes.
- Click Auto-Detect Walls. Walls, doors and windows appear as normal Strucit objects on top of the image.
- Fix what’s off. Drag wall ends, delete extras, add what’s missing. Then hide the image or keep it underneath as a reference.
Drawing styles it reads
Strucit doesn’t use one detector for everything. It looks at the pixels and picks a model for the drawing style:
| Style | What it looks like | Doors and windows |
|---|---|---|
| Solid walls | CAD exports and scans where each wall is a dark band | Yes |
| Hollow walls | Each wall drawn as two thin lines with white in between | Yes |
| Marker sketch | Phone photo of a thick, re-traced marker drawing, uneven light | Yes |
| Pen sketch | Phone photo of a thin freehand pen or pencil plan, rooms labeled by hand | Walls only |
TP-Link Omada floor-plan reports have their own importer: see Omada import.
Runs in your browser
Detection runs on your device, in a background worker. Your plan isn’t sent to an AI service, which matters when the drawing is a client’s building. It also means it works on a plan you can’t or shouldn’t upload anywhere.
Free and Pro
Wall detection is a Pro feature. On the free plan you can still upload the image as a reference and trace the walls by hand with the wall tool.
What it doesn’t do
- It reads the drawing as pixels: a PDF is rendered one page at a time, and vector layers aren’t read as CAD geometry.
- It doesn’t read room names, dimensions or text.
- One image is one floor.
- Photos taken at a steep angle, heavy shadows and walls that don’t meet lower the accuracy. Straight-on, evenly lit photos work best.