Precision crop load management, or PCLM, is a is a scientific approach to calculating optimal crop loads for fruit crops in order generate maximum year-over-year profits. This is accomplished by pruning and thinning throughout the growing season, reducing surplus buds, blooms, and fruit to enable the tree to produce the most high quality fruit while preserving its ability to bear fruit the following year.
The Promise of PCLM
Dr. Terrence Robinson, a fruit physiologist from Cornell and one of PCLM’s originators, had presented long-term research that demonstrates PCLM has the potential to deliver between $5 and $10 per tree per year in increased productivity. That’s a big incentive to drive adoption.
The Challenge of PCLM
However, while apple growers almost universally embrace the ideas behind PCLM, and indeed many of them practice some sort of crop load management, the practice has yet to be fully embraced. And Dr. Robinson has an idea why that is.
“The best available precision crop load management strategies today are tedious,” said Dr. Robinson, in a March 2021 article on crop load management in Good Fruit Grower, adding in another article, this time from Fruit Grower News (March, 2020), “Computer visioning is a technology that would revolutionize the process of counting buds.”
In our own experience talking to growers, hand counting and sizing is time-intensive, error-prone, and hard to verify. Growers end up sampling a small — arguably, a statistically insignificant — number of trees and fruit. That number rarely breaks a tenth of a percent (0.1%) of the total trees they manage and is more likely to come in at about a quarter of that number, or 0.03%.
An Explosion of Computer Vision Offerings
So it is no surprise there has been a recent uptick in companies offering hi-tech solutions to solve the counting problem. Most of these companies offer tractor-mounted high-end cameras and are used principally to count fruit. A handful of companies, FruitScout included, have gone the mobile phone route.
Mobile phones may lack the advanced image capture capabilities of their $50,000 tractor-mounted counterparts, but there cameras are still quite good. The latest phones capture 21 megapixel images. That’s more than double the capability of top end phones offered just five years ago and more than enough for computer vision use cases.
Mobile vs. Dedicated Camera for Data Collection
More importantly, mobile phones don’t cost extra. Everyone already has one in their pocket and knows how to use them. And, unlike the tractor-mounted cameras, they can be used for both counting and sizing of trees of any geometry — single spindle, v-trellis, dense rows or sparse ones.
The biggest disadvantage of mobile phone-based data collection is poor connectivity in the orchard. And the fact is, tractors can move faster than a person walking a row taking pictures of all the trees. Still, at FruitScout we’ve seen collection rates of 200 trees in less than ten minutes. For a 5,000 tree block, that’s 25 minutes to grab a 10% sample rate.
And 10% sample rates lead to much more accurate crop load targets and better financial outcomes. At least, that’s the plan.