Drones and machine learning are used to identify male plants in hemp seed production fields. Inspectors currently visit fields twice per season and manually count about 10,000 plants to determine how many are male, which is physically demanding in tall fibre hemp crops. Drone imagery from hemp breeding plots enabled recognition of visual differences between male and female plants. A machine-learning model developed with engineers and programmers distinguishes plant structure and appearance using vision recognition rather than multispectral analysis. The system is in its third year and could be adapted for other crop inspection and agronomic applications beyond hemp. Canada’s hemp sector is also expanding through developing fibre processing capacity in Alberta and regulatory consultations to reduce red tape for industrial hemp production and processing.
"In seed production... inspectors go to the field twice a season, count 10,000 plants, and count how many of those plants are male plants. Slaski says, noting the work is physically demanding in fibre hemp crops that can reach several metres tall."
"I looked at the pictures and I said, 'Wait a minute... pictures are just good enough taken by the drone to (distinguish) between male and female plants,' Slaski says."
"Rather than relying on multispectral analysis, the system uses vision recognition and machine learning to identify differences in plant structure and appearance. Slaski says the technology could eventually be adapted for other crop inspection and agronomic uses beyond hemp."
"The project, now in its third year, grew out of work with InnoTech Alberta engineers and programmers to develop a machine-learning model capable of distinguishing male and female hemp plants."
Read at Realagriculture
Unable to calculate read time
Collection
[
|
...
]