Cart

Overview

Drone systems can be very effective for monitoring trial plot sites. This is because they can fly low to the crop, getting a high level of detail, satellite imagery is simply not high enough resolution to see individual plots.

This season, we have focused on finding out what data and results we can get using a simple, off-the-shelf drone system with a standard RGB camera. Using a combination of mapping software and Skippy Scout to collect super high resolution imagery of the plots.

While the mapping provided a good overview of which plots were establishing well, the standard camera did not help so much in identifying disease at later stages. Skippy Scout was able to collect imagery that allows enough detail to look at individual plants.

Comparing results to last year, we can see that this years crop is a lot thinner due to the weather, exposing the ground and dead plant matter even at later growth stages.

Data was collected at four key times throughout the season, to look at early establishment and later crop progress and possible disease detection.

A DJI Mavic 2 drone system was used to collect all the data, alongside software programs Pix4D Fields, Dronedeploy and Skippy Scout.

Plots were mapped and the data processed with the VARI index only, as this can be used with standard off-the-shelf cameras. VARI essentially helps to highlight where there is more growth/plant matter, and where there is less, it looks at ‘greeness’ and so in the right conditions can also highlight an unhealthy crop where leaf coloration has changed.

We also used Skippy Scout, a piece of software that automates drone flight to capture images close to the crop, with leaf level detail. For trial plots this can be useful to get an idea of what is happening to individual plants, and to ‘ground truth’ the higher level drone mapping data.

Project

MountFair Wheat Trials 2020

Drone System

Mavic 2 Pro

Software

Pix4D, Dronedeploy, Skippy Scout

October 30th

3 weeks after planting
Low Growth
Elation
High Growth
KWS Barrel
  • Above, Skippy Scouts plant detection AI gives a close up of two example plots.
  • In the map to the right, green areas show where the crop is starting to establish. Red areas show less growth.
High Growth
Low Growth

November 20th

7 weeks after planting
Low Growth
Elation
High Growth
KWS Barrel
  • We can see 3 weeks later that results are similar in terms of which varieties are establishing faster.
  • In the map we use a TGI index to see if there any any differences, however results are the same.
High Growth
Low Growth

May 6th

31 weeks after planting
  • Map and VARI Index still shows variation across the plots.
  • This season weather has had a large impact, growth is thin and those that established well early seem to be do better now.
  • Due to the amount of ground and dead plant matter viable, disease variation cannot be identified, even though present.
Low Growth

June 5th

35 weeks after planting
  • The AI image analysis failed in the later stage crop to detect disease. We have some work to do!
  • Comparing results to last year, we can see that this years crop is a lot thinner, and further along, due to the weather this season.
  • This exposes the ground and dead plant matter even at later growth stages, which is confusing the disease detection systems.
Taken June 4th 2019 – Previous season
Disease detected
June 5th – This season
Thin crop

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.