Quick results or full analysis — pick the workflow that fits your needs.
Upload a street photo and get an instant pedestrian-friendly redesign. No prompts, no instructions — the pre-trained model runs automatically.
Complete human-in-the-loop pipeline: grid-based area analysis, pedestrian-vehicle interference detection, ML generation with SUMO simulation, and LLM agent for iterative review.
From detection to engineering drawings — a complete pedestrian infrastructure pipeline.
Upload street photos with EXIF geolocation, enter lat/lng coordinates, or tap an interactive map. The system auto-fetches OSM, Street View, and satellite data.
YOLOv9 detects encroachments, potholes, open drains, and broken sidewalks. LLaVA understands land use, street character, and infrastructure problems.
Pedestrian infrastructure, safety, walkability, monsoon resilience, accessibility, green coverage — each scored 0–100 against IRC:SP:103 and Vision Zero.
Deterministic engine enforces 6 hard rule categories: continuous sidewalks, frequent crossings, road diet, tree canopy, covered drainage, and speed management.
CRITICAL → LONG-TERM interventions with specific locations, cost estimates, and per-meter BOQ breakdowns.
GeoJSON, LandXML (Civil 3D), DWG (AutoCAD), SUMO simulation, PDF reports, before/after renders, and interactive Leaflet maps.
Photo upload, coordinates, or map selection
OSM, Street View, Sentinel-2, OSRM routing
YOLOv9 + LLaVA + EasyOCR + SegFormer
7 dimensions against international benchmarks
Constraint-validated, IRC-compliant
8 formats ready for civil engineers
UrbanX compares your streets against Copenhagen, Amsterdam, Tokyo, Barcelona, Singapore, and Seoul.
Zero cost. Zero training. Just drop a pin and let UrbanX do the engineering.