Plant row counting fail. Farmer 1 Drone 0

I have been waiting for the weather to play ball. I want to see if some of the drone agriculture products out there really do what they say they will do.

Speaking with a farmer friend of mine a couple of weeks ago he told me his recently planted seeds would be out of the ground and ready to be counted last week.

South Africa has been experiencing one of its worse droughts ever, but now I want it to fly of course it has been windy, raining or misty or all three at once.

Yesterday it cleared for a few hours and I popped up the road to give it a crack.

Quite by chance, my friend Steve had taken just the same gap and was walking the field counting plants.

Precision agriculture drones are a thing, not a week goes by without me posting something about them. Can it all be true? Time to chalk up a win for the drone!

My plan was to fly an area of about 11 Hectares at a low enough altitude that would allow a resolution of at least 2.5 cm a pixel. This is a requirement of DataMapper from PrecisionHawk and its counting algorithm.

I actually planned the flight in DroneDeploy’s app, quite simply because of all of them out there I like it the most. Bonus feature (as they all are) its free to use.

As always the area seen is larger than planned because of the overlap all around. In this case, it ended up being around 17 hectares flown. At a resolution, so the software tells me of 2.35cm a pixel. The drone took 380 photos.

As soon as I arrived on site I realised my journey might have been in vain. The plants had only just popped out of the ground. They were between 2 and 5cm in length and just single stalks. A very big ask for a camera to see them from 50m away.

I can’t see one plant with the naked eye in the imagery.

So, for now, Steve wins, although in the end, I should be able to cover far more ground than him and his clicker.

I’m not going to fork out the $9 to have DataMapper process this one, I just don’t think row-based-counting-priceits worth trying. That $9 covers a months worth of processing, $90 would get you a year. Waiting until I can see the plants easily in the images should ensure success and give me a chance of two surveys over a month for $9.

I have learnt lots of things, though.

I know for certain that my chosen flight area is free from obstructions. You can’t see that on flight planning apps.

I know that my Phantom 3 Professional can easily fly the task so I can go lower. It used only 55% of its battery so another metre or two lower will be ok. This will increase the resolution slightly which is always a good thing.

So a reasonable start, I shall return next week to try again.

With the right sensors on a cheap platform and enough friction removed from gathering data, drones might yet be useful down on the farm.