Steve Becker admires the Inspire UAV primarily used by other people to fly over his corn breeding program plots. The Beck’s plant breeder, one of six plant breeders pursing their own development programs, understands that flexibility of this particular unmanned aerial vehicle allows him to get somewhat better images than those taken with other UAVs, although those images are fine for their use. The Inspire model allows for more sensors and different camera angles, which can result in clearer pictures.
Becker is way past the “gee-whiz” stage when it comes to drones. It’s not really the drone that excites him, but what he can do with images from the drone. Plant breeders of 50, 40, maybe even 10 years ago could only dream what it might be like to get an image of breeding plots from overhead in enough detail to help make decisions. And Becker says this is only the beginning.
What’s possible now
The secret is working with companies that take data from the UAV flights and conduct detailed analysis of what the images show, Becker says. He continues to work one on one with software developers who know what he needs, and are confident they can deliver it. So far he’s getting data such as stand counts and plant height on his corn plots. Previously, technicians had to manually take that data.
“It frees them up to move on to recording ear heights and other traits,” Becker says.
But what analytics can do is only in its infancy, Becker says. Within as little as two years, the software may well be able to take the UAV images and determine ear height, lodging, disease scores and other important characteristics, as well.
What future holds
The really exciting potential opportunities become apparent when Becker pulls up drone images and zooms in on disease lesions on leaves. “Artificial intelligence is coming for these applications,” he says. “In fact, it’s coming within a relatively short time frame.”
He explains that he will be able to train the system to pick out what he is looking for. In other words, artificial intelligence can learn what to look for in images. If Becker wants it to find gray leaf spot lesions on leaves, he can train it to recognize gray leaf spot lesions. Then the system will be able to zoom in and identify plants within the breeding program with more gray leaf spot compared to other plants.
“It’s not only the time savings that will help,” Becker says. “Right now I can’t visually see all of my plots all the time; it’s physically impossible. When these systems are perfected, we will not only be able to scout them all and do analytical analysis, but we will be able to save those records and compare year to year.”
Will it mean turning out new corn hybrids faster? Not necessarily, Becker says. But it will mean being more precise and improving a plant breeder’s ability to select the very best germplasm available, and not overlook potential sources.