Wild pine within dense green vegetation viewed from a drone

Case study 01 · Wild pine

Wild pine detection
using drones and AI.

Using high-resolution drone imagery, human analysis and AI to find and locate visible wild pine trees above dense scrub canopy.

Read the case study ↓
Conditions
Overcast · no shadows
Image detail
≤0.3 cm GSD
Review
Human + AI

The issue

Finding wild pines
in impenetrable scrub.

Finding wild pines across dense, impenetrable scrub is difficult.

The thick scrub canopy limits visibility. The detectable trees are the wild pines visibly poking their heads above the surrounding canopy.

Wild pine example visible in detailed aerial vegetation imagery
Training exampleWild pine visible among surrounding vegetation

The solution

High-resolution imagery.
Human + AI analysis.

Drones capture high-resolution imagery across the terrain. Flights aim for cloud cover or overcast conditions with no shadows, and a ground sample distance of 0.3 cm per pixel or finer. The imagery is then processed and analysed by people and AI to find and locate all visible wild pine trees poking above the canopy.

  1. 01

    Capture for detection

    The flight is weather-dependent. We aim for cloud cover or overcast conditions so the canopy is shadow-free, giving the imagery the best detection potential. The target ground sample distance is 0.3 cm per pixel or finer so wild pines can be distinguished from similar-looking vegetation.

  2. 02

    Reconstruct

    The individual photographs are stitched into one detailed, georeferenced aerial view so the whole search area can be reviewed in context.

  3. 03

    Train

    Clear wild pine examples from that job are labelled alongside surrounding vegetation. The model learns from the imagery and conditions found on the site.

  4. 04

    Detect + verify

    AI searches the reconstruction for likely targets. An operator checks every candidate and keeps only detections that can be approved with confidence.

  5. 05

    Export

    Approved locations become a clean target layer that can be used to plan the most appropriate inspection or treatment method for the job.

Detailed georeferenced aerial reconstruction used for AI-assisted vegetation review
Working mapOne detailed reconstruction gives the search a consistent spatial reference.

Why the process matters

AI narrows the search.
A person makes the call.

The model proposes likely wild pine locations across the mapped area. Each candidate is then reviewed against the original imagery. Only approved detections are carried into the final dataset.

Detection boundary · Field evidence

Where aerial detection
stops.

Drone imagery can locate wild pines that are visible above the mānuka canopy. Pines growing underneath the canopy cannot be seen from the air. Finding those concealed trees still requires someone to physically push through the dense scrub and inspect it at ground level.

Aerial viewAn example of what can be found from aerial imagery.
Ground truthField footage reveals a pine concealed underneath the mānuka canopy.
Small wild pine growing underneath dense mānuka scrub
Hidden pine 01From above, the surrounding canopy blocks this tree from view.
Young wild pine concealed among dense dark scrub
Hidden pine 02Locating concealed trees like this requires a physical ground search.

The outcome

A clearer path
from image to action.

01

A mapped site

A high-resolution aerial reconstruction that preserves the wider property context.

02

Verified targets

Wild pine candidates reviewed by a person before entering the final layer.

03

Actionable data

Approved locations prepared for the most appropriate follow-up or application workflow.

The exact capture plan, review threshold and export format are matched to the site and the intended follow-up work.

Aerial vegetation prepared for weed detection review

Your property

Would you like a hand
to find wild pines?

Tell us about the target, terrain and area you need searched. We will help determine whether AI-assisted drone mapping is the right fit.

Discuss a detection project ↗