Northline — Vegetation Health

Drone Multispectral · Sample Deliverable

VEGETATION HEALTH, FROM ONE FLIGHT

A drone with a multispectral camera sees plant stress the eye can't. One pass over this maize research field becomes a health map of every plot — processed end-to-end on Northline-controlled infrastructure in North America.

Hover the field

The whole field, graded

Green = thriving canopy. Yellow/red = weak or bare. Hover any point for its value; drag the divider in 3-week change to see the crop establish.

bare / stressedthriving

Weak-canopy finder

Flags weak canopy on the Vigour map.

Vigour threshold0.35

Field imaged—

Canopy cover—

Canopy flagged weak—

Area box

In Area box mode, drag a rectangle around any plot for its own numbers.

Switch to Area box and drag on the field.

Season, tracked

Sampled plot tiles across four flights.

Whole field, May 25 → Jun 15—

PLOT-LEVEL DETAIL

Zoomed to individual plots: the same maps at leaf scale. The photo shows green; the maps show which green is healthy — and the analysis counts every plant.

HOW IT WORKS

01Fly

Your site is captured in a single multispectral drone pass — ours, or a vetted local pilot flying to our spec.

02Process

Imagery is processed into calibrated health maps on Northline-controlled infrastructure in North America — never handed to a third-party scanning service.

03Review

Maps arrive in your Northline portal with expert analysis — measurable evidence behind every recommendation.

These detections are our own model's output, not the dataset's labels. Northline vision model v1 (RF-DETR), measured on held-out test imagery: maize AP 0.56; the weed classes score 0.196, 0.144 and 0.035 and under-detect weeds badly — on this dataset the expert labels carry 70–235 weeds in frames where the model finds a handful. Counts are shown for illustration; no weed-pressure judgement should be drawn from them, and a grower's decision needs the trained-up model and an agronomist. Health maps are scan-derived evidence to support expert review — not a laboratory measurement. Sample imagery: WeedsGalore research dataset, Çelikkan et al. (GFZ), CC-BY-4.0 — shown to demonstrate processing. © Northline Readiness.