In the city of Arusha in Tanzania, David Guerena, agricultural scientist at the International Center for Tropical Agriculture, is leading a project called Artemis.
Funded by the Bill and Melinda Gates Foundation, this is using AI to help breed more resilient crops. Specifically the AI is helping speed up work called phenotyping.
This is the visual studying of new crop varieties based on observations of their characteristics, such as how many flowers, pods or leaves that a plant has.
“Traditionally it takes around 10 years to develop a new crop variety,” explains Mr Guerena. “But given the pace of climate change, this timeframe is no longer viable.”
He adds that the phenotyping work traditionally relied on the human eye. “But humans are just not doing this consistently, with the high levels of precision necessary, to make subtle, yet important, plant selections,” says Mr Guerena.
“It can be over 30˚C in the field. It’s just tiring, and fatigue affects data quality.”
Instead, growers involved in the project are taking photos of their crops through an app on a smartphone. The trained AI can then quickly analyses, records, and reports what it sees.
“Computers can count every flower or pod, from every plant, every day without getting tired,” says Mr Guerena. “This is really important as the number of flowers in bean plants correlate to the number of pods which directly influence yields.
“Data can be so complicated, to understand what’s happening, but AI can be used to make sense of that complicated data and pick up patterns, show where we need resources, show recommendations.
“Our plant breeders estimate that with the better data from the AI computer vision they may be able to shorten the breeding cycle to only a few years.”
