CanopyOS: Open-Weight Vision for Outdoor Urban Greening Observations

Description and use

CanopyOS is a mobile-first prototype that uses an open-weight segmentation model to generate transparent reports on urban greenery. It processes photos locally, computes coverage and a Green Score, and offers cautious recommendations without replacing field measurements.

Problem

Can an open-weight vision model support a transparent first-pass observation of an outdoor scene while making the limits of that observation visible?

Outcome

The prototype successfully generated traceable reports for two openly licensed photographs, demonstrating that local inference can produce a Green Score and recommendation without sending data to a hosted AI provider.

Workflow

• Take a photograph of an outdoor scene
• Upload or capture image in the app
• Run local inference with nvidia/segformer-b0-finetuned-ade-512-512
• Map ADE20K labels to vegetation, hard surfaces, buildings, and open ground
• Compute classified coverage and Green Score using the disclosed heuristic
• Display segmentation overlay, indicators, and recommendation in a report
• Store scan history privately and allow deletion

Evidence

CanopyOS runs the pinned nvidia/segformer-b0-finetuned-ade-512-512 model locally after its weights are downloaded.

Source

The report separately shows classified coverage, the fraction of the whole image represented by those four groups.

Source

Green Score = round(100 × (0.60 × vegetation_fraction + 0.25 × open_ground_fraction + 0.15 × tree_presence))

Source

The software validation record reports 30 passing Python unit/API tests and 12 passing Playwright desktop/mobile tests.

Source

The next evaluation is to photograph multiple current outdoor locations with permission, manually review their masks and recommendations, and record failure cases.

Source