Where we started, where we are
Every gain below was made on free data alone. No labels bought, no third parties waited on. All figures MEASURED this run.
The classifier's ceiling
How separable are the 14 material classes under the model's own best case, and how far can cleaner pixels push that ceiling?
Separate vs hybrid reflectance
Measured on 184 real roofs. Physics (NTB) is material-independent; the lookup table (LUT) inherits every classification error.
The resolution wall
On real UK building geometry: how many roofs yield a clean pixel at each ground sample distance, and which sensors carry the SWIR albedo needs.
The sourcing cabinet
Imagery, footprints and methods, researched and cited. Recommended rows carry a blue margin.
The free frontier
Three free sources and techniques we had not used, pulled live and sampled at the same real roofs. How far can we get today, with no labels, no paid imagery and no waiting on anyone?
Closing the loop
The deepest fix, run today: harvest real roof-material labels free from OpenStreetMap, train on those instead of textbook spectra, and fuse the free radar. The first measured accuracy this project has had.
Inching toward quality
The lane-separated architecture and the ordered moves that convert this prototype into a defensible product.
To Silicon-Valley tier
Beyond fixing the prototype, the moves that make this a category-defining product, not a script.
Glossary & sources
Every term, data source and model on this page in plain English, enough to explain what we do to someone new.