05·3D & Spatial
Scan to Floor Plan
A scan already knows the shape of a building, but a buyer wants a floor plan: walls, rooms and doors, drawn to scale on one page. Six generations of pipeline tried to get there, ending in a machine-learning model built on published architectural-reconstruction research. The version that ships is not that one.
ResultThe plans published on this site were drawn by the ninth version, a plain grid method, and their title block says so.
[ Ask about this build ]01Discover
Why it exists.
A scan carries the building's true geometry, and almost nobody wants to look at a point cloud. What sells a property is the drawing every buyer already knows how to read: a plan, to scale, with the rooms named and the doors in the right places. Between those two things sits the actual problem, which is deciding what counts as a wall.
That question is much harder than it sounds, because a scan of a lived-in home is full of things that look like walls and are not. Wardrobes, curtains, kitchen islands and half-height partitions all read as vertical surfaces, and a plan that promotes any of them to architecture is wrong in a way a client will spot immediately.
02Define
The brief.
Turn a 3D scan of a real property into a floor plan a buyer can read: walls, rooms and doors, drawn to scale on a single page.
03Develop
What it took
Skills behind it.
Primary discipline plus the support stack.
Skills demonstrated
- ResearchPrimary
Scored six approaches against each other on the same eighteen scans rather than assuming the newest was best. The comparison is the deliverable: without it the sophisticated pipeline would have shipped on the strength of being sophisticated.
- AI & Machine LearningPrimary
Built the learned pipeline properly before judging it, including auto-labelling the training scans, two rounds of fine-tuning and a scoring overhaul, so its loss could not be written off as a half-hearted implementation.
- 3D & SpatialSecondary
Diagnosed the loss rather than accepting it: the model learns from synthetic buildings and is asked to draw real scanned ones, and it consistently under-covers the true area, which costs it more than accurate room detection wins back.
- Data PipelinesSupporting
Kept every generation runnable behind one orchestrator, so the comparison could be re-run on new scans and the method that ships stays whichever one actually scores best.
Develop · Build
How it came together.
- 01
Six generations, kept side by side
The pipeline went through six approaches, from fitting geometry directly to the point cloud, through grid and building-footprint methods, to a learned model built on published architectural-reconstruction research. Each new generation stayed runnable rather than replacing the last, which is the decision the rest of this depends on: keeping the old ones alive is what made it possible to find out that a newer one was worse.
Data Pipelines - 02
Score them against each other, on the same scans
Every generation was scored on the same eighteen scans rather than demoed on the ones it handled well. That sounds obvious and is the step most often skipped, because the newest approach usually arrives with a story about why it is better and a couple of examples that support it. A shared scoreboard is what turns that story into a claim that can be wrong.
Research - 03
The learned model lost, and kept losing
The machine-learning pipeline is the most sophisticated thing here and it wins on two scans out of eighteen. It was not dismissed on a first result: the training scans were auto-labelled, it went through two rounds of fine-tuning, the scoring was overhauled, and the output polygons were expanded. Its scores improved substantially and the win rate did not move. Being beaten after a fair fight is a much more useful result than being beaten because nobody tried.
AI & Machine Learning - 04
Why it loses, which is the part worth keeping
The model learns from synthetic buildings and is asked to draw real scanned ones, and on real scans its rooms come out consistently smaller than the rooms actually are, by a margin large enough to lose on total area even where it identifies the layout correctly. That is a gap between where the training data comes from and where the work happens, not a tuning problem, which is why more fine-tuning kept not fixing it and why the next thing to try is combining it with a method that gets the outer shape right.
3D & Spatial
04Deliver
What shipped.
The plans published on this site were drawn by the ninth version, a plain grid method, and their title block says so. The machine-learning pipeline that superseded it on paper wins on two scans out of eighteen, and still won two after fine-tuning. Knowing why is worth more than the pipeline would have been.
The floor plans published on this site were drawn by the ninth version, a plain grid method, and their own title block says so. Anyone can check: the label reads grid_v9. That is the honest state of this work, and shipping the version that scores best rather than the version that sounds best is the whole point of having built the scoreboard. The learned pipeline is not abandoned; it is waiting on being paired with something that fixes the area shortfall, which is a different piece of work from tuning it harder.
By the numbers
Generations built
geometry through to ML
ML win rate
unchanged after tuning
Area shortfall
why it loses
See also
3D & Spatial
Spatial Asset Intelligence
Turning one handheld building scan into a fire-safety asset register somebody can check, and deciding which classes were dependable enough to hand over.
3D & Spatial
3DGS Research Pipeline
A reproducible harness with 26 Gaussian-splat engines wired into it, scoring which reconstruction survives a real luxury interior.
3D & Spatial
Barnes Vantage
An international portal for viewing real Dubai residences in 3D: walk a real scan yourself, or have a broker drive the same walkthrough live.