Nabhya reconstructs near-NDVI crop health maps from standard RGB drone footage — no multispectral sensor, no hardware changes.

A multispectral camera costs ₹8–12 lakh. Most Indian drone operators — 1,400+ DGCA-certified — fly standard RGB sensors and generate zero crop health data per flight.

Actual model output — unedited.
Every output cross-checked against held-out Sentinel-2 ground truth. SSIM 0.8060 across 440 validation images.
Developed to make satellite-level crop analytics affordable to all growers in developing ecosystems — bypassing multi-lakh camera hardware.
SSIM 0.8060 on 440 held-out Sentinel-2 images — a reproducible, auditable methodology. 16.8% above the published benchmark.
Under 2 seconds per image. Designed for batch processing — submit a full flight's worth of frames in one API call.
Built for the Indian agri-drone ecosystem. Trained on Indian field imagery.
Every score tied to ground-truth imagery. Exportable JSON/PNG — designed for PMFBY-claims documentation.
MVP — field trials in progress. Early partners get direct access to the founding team.
1st Prize, AgriTech Yuva Competition 2026 — selected among 11 national finalist teams.
Pilots run on your existing flight imagery — no hardware changes, no fleet disruption. One REST API call returns JSON + PNG output.
Galgotias University, Greater Noida · AgriTech Yuva 2026, 1st Prize · IEEE Hackathon Winner 2026
No signup required · under 2 seconds · actual model output
Open Live Demo →Interested in partnerships, pilots, or demos? Let's talk about your farm network and monitoring needs.