A Western study could help farmers get out of a potential jam by using artificial intelligence (AI) and passive camera monitoring to enhance strawberry cultivation.
In a paper published in the international journal Foods, Western engineers describe a new machine-learning approach that yields the highest-ever precision and accuracy rates for ripeness and disease detection in strawberries of any previous attempts.
AI and computer vision have been used to monitor a variety of different crops over the past two decades and most of the time the surveillance is done by highly expensive, third-party, private companies taking control out of the hands of farmers.
Joshua Pearce, the John M. Thompson Chair in Information Technology and Innovation at Western Engineering and Ivey Business School; Soodeh Nikan, a Western electrical and computer engineering professor; and their collaborators are working on a new project, funded and supported by the Weston Family Foundation's Homegrown Innovation Challenge, to greatly extend the berry growing season in Canada.
And their approach is available to everyone. For free.
"One of the approaches we are taking to drive down future costs is using computer vision and automated detection," said Pearce. "As always, we took an open-source approach and integrated the latest and greatest techniques to substantially improve the overall precision and accuracy."
While future studies will use the now-proven model for traditional outdoor crops, this project was executed in an agrovoltaic agrotunnel, an indoor growing system that houses high-density vertical aeroponic (growing plants in the air) and hydroponic (growing plants in water) hybrid systems that use high-efficiency, spectrally optimized LED grow lights. All systems on the inside are powered by agrivoltaics (solar photovoltaic systems over crops) on the outside.
"AI has been used before in crop monitoring, but nothing close to the precision we have achieved. We have greatly increased accuracy in detecting different diseases and also sensing the brightness of the strawberries, which is crucial for understanding the quality of the crop and determining the best times to pick," said Nikan.
The new model approached nearly 99 per cent accuracy in terms of strawberry ripeness and disease classification, making it a potentially valuable tool in the global crisis of food waste. Its real-time performance is critically important, as it allows for continuous and accurate assessment of the crop throughout the growing season.
Joshua Pearce and Soodeh Nikan investigate strawberries grown under an agrivoltaic installation at the Environmental Sciences Western Field Station. (Jeff Renaud/Western Communications)
Yield futures, not futuristic
AI can detect with high accuracy, but it often requires a giant dataset that must be stored on a computer with a large amount of memory. It's also costly.
With agrivoltaics/agrotunnel operators and small to mid-size farms in mind, Pearce and Nikan were focused on finding a solution for everyone in the agriculture sector.
"We wanted to reduce the size of these AI models to make it something feasible for farmers and localized production," said Pearce. "We didn't want to just increase the accuracy, which is above 98 per cent, but also reduce the size of the models."
For the study, Pearce, an expert in solar-powered, open-source 3-D printing and recycling that uses AI for quality control, teamed up with Nikan, who has years of experience working in AI and computer vision on everything from self-driving cars and screening for hearing disabilities to sleep patterns and heart arrhythmia.
"Reducing waste and the cost of food is obviously a big issue these days. Like everyone, I am always surprised when I go to grocery store and see the price of fresh fruits and vegetables," said Nikan. "When choosing projects, I usually look for something that is safety critical or a societal need. With my experience in other applications, I jumped at the chance to apply my knowledge and expertise to food security."
The new Western AI model only requires a small amount of initial data, in this case images, to generate a much larger working dataset for ripeness tracking and disease detection in strawberries.
And because the new model is completely open-source, it puts all farms on an even playing field - whether they are big or small or grow inside or outside.
"The software is completely free and open-source and farmers of any type are free to download it and then adapt it to their needs," said Pearce. "They may prefer to have the AI system send them an email or ping their phone when they detect disease or even forward an image of a specific plant that is ready to pick. The software is wide open to make it your own."
For the next step, Pearce and Nikan are looking to experiment with the software outdoors, possibly using drones, to monitor more traditional strawberry fields. In addition, they are using AI computer-generated synthetic images to further cut down on the number of real photographs required.
"As opposed to taking images of millions of strawberries, which is a low efficiency, high-cost approach, we are using synthetic images and open-source software to create millions of images ourselves, with relatively low computer power, which now allows us to pinpoint highly granular observations about ripeness and disease for very specific plants," said Nikan.
