14 Ate Today? Thank a Farmer. Better Farming | December 2024 Research & Farm Science mentor and researcher, Prof. Alireza Navabi. “I came into a supportive environment when there was a gap in leadership, and people really came together,” Booker says. “Now my role is to steward the program, and then lead it onto the next generation.” AI solutions point to future of wheat breeding The next generation has already begun the work. Thanks to two of Booker’s graduate students, Riley McConachie and Connor Belot, a free app will allow farmers to take a photo of their wheat plot using their smartphones and receive a near-instant analysis of wheat head infection levels. “Riley’s research has built the technology that allows you to take a picture of a wheat plot and evaluate the disease,” Belot says. “My research asks: How do you scale that up to the field level? That involves developing some way to use satellites or drones to get a field-level average.” “It’s worked out better than I think anybody would have expected,” McConachie says. He adds the app is about 86 per cent accurate on the field level, improving with better technology and deep learning capabilities every year. Assessing FHB manually – that is, standing on the field and trying to visually determine its severity over hundreds of wheat heads – is time- consuming, subjective and inaccurate. Many foliar (leaf) disease signs include hard-to-spot brown or orange discolourations, which could be easily confused with things like hail damage. The app could prevent those costly human errors. By meticulously labelling field images of wheat plots, McConachie and Belot trained one open-source, deep-learning technology to create a perfect outline of the wheat head. U of G students get ready for future of agriculture The artificial intelligence (AI) model students trained, YOLOv8, identifies the general whereabouts of the wheat head. They incorporate it with another model, Segment Anything by Meta AI, that uses this general location and traces an accurate image of the head that filters out any background information. From there, the app applies a spectral index. This essentially uses the different red, blue and green values captured by a cellphone to determine if the wheat head is diseased, and by how much. As McConachie describes, integrating these AI models with spectral index has not been done before. The two worked closely with Dr. John Sulik, plant agriculture professor at U of G, who they say was critical in helping develop and test the methods. “Spectral index isn’t anything new, but people kind of forgot you could use it like this,” McConachie says. “We brought back something that was on the fringe, concepts developed by NASA in the ’70s, and then we combined that with deep learning to create a completely new approach.” The two hope the app will be ready as early as next year, bringing farmers yet another powerful solution to FHB. In the face of climate change and food insecurity, Booker and her students continue their essential work to help put bread on the table. Along the way, hands-on experience provided by Booker and her team of researchers helps students like McConachie and Belot refine their industry knowledge, learn new methods and gain valuable leadership and problem-solving skills to succeed as part of Ontario’s agri-food workforce. And though the battle between wheat and fungus rages on, with every new seed and technological innovation, farmers may just be ready for the fight. BF Dr. Helen Booker hopes to end wheat contamination. University of Guelph photo
RkJQdWJsaXNoZXIy NTc0MDI3