Acts like the external tool baseline so we can visually benchmark class coverage and scene interpretation.
AI-Powered
Compare a reference segmenter against the Urban Leaf segmentation workflow.
Upload one image, run both engines on the same frame, and inspect masks, overlays, class ratios, and canopy-style indicators in one place.
Pre-processed results
Uses the project class system and indicator extraction so the website can explain our scene understanding.
The UI contract is ready today and can later swap this project engine for the real checkpoint inference path.
Input frame
Upload an image to see results here…
Agreement…
Canopy estimate…
Water / shadow…
Reference segmenter
Baseline output for side-by-side comparison.
Urban Leaf workflow
Project-oriented segmentation and indicator extraction.
Derived scene indicators
Quick signals we can surface beside segmentation output on the website.
Vegetation plus sparse vegetation share from the project segmentation.
Urban or impervious class share inside the uploaded scene.
Bare soil plus built-up pixels, useful for disturbance-style interpretation.
Low-light and water-like regions that influence scene readability.