The wild blueberry market has grown into a multi-million dollar industry in North America, with consumers pushing up demand as they seek out the prized berries for a range of health benefits.
However, growing and harvesting the fruit can be complicated by weeds that infest commercial fields, compete with blueberry plants for nutrients, and interfere with harvesting equipment. To stem the spread of unwanted plants, farmers often rely on widespread applications of herbicides to control weeds, some of which have become resistant to chemicals through repeated use.
Dalhousie University’s Mechanized Systems Research team is tackling this problem through the use of deep learning convolutional neural networks (CNNs) – an emerging smart technology that can identify weeds and empower farmers the information they need to apply agrochemicals in a more targeted manner.
Travis Esau, Assistant Professor in the Engineering Department of the Faculty of Agriculture, recently published an article on this topic with his doctorate. student, Patrick Hennessy. This research determined that the technology could detect sheep sorrel and hair fescue – two common weeds that affect Nova Scotia’s blueberry crops – in images of nearly 60 fields of the province’s wild blueberries. .
“Through this research, we discovered how CNNs can very accurately detect the presence of hair fescue and sheep sorrel when applying herbicides,” said Dr Esau.
âAs a result, we can integrate the trained CNN into smart applicators to enable spot treatment of agrochemicals and prevent significant amounts of inputs from being unnecessarily wasted, resulting in more sustainable and respectful management decisions. of the environment.”
How it works
CNNs can intelligently identify visual features and find patterns associated with the target plant with minimal user input, making them easily adaptable to new targets. They are trained to detect new targets by repeatedly showing a computer numerous labeled images of the desired target, in this case particular weeds.
The technology was used to indicate the stages of maturity of blueberries and estimate potential yields. CNNs have also effectively detected weeds interfering with strawberries, potatoes, Florida vegetables, and other crops. They have been used to detect diseases on tomatoes, apples, strawberries and various other plants.
Eight digital cameras captured thousands of color images of fields of wild blueberries containing fur fescue and sheep sorrel in central and northern Nova Scotia. CNNs trained with these images identified hair fescue and sheep sorrel with accuracies of up to 97% and 90%, respectively. The results of this study indicate that CNNs provide a high level of accuracy for weed identification in wild blueberry fields.
Dr Esau says the technology could make farming much more efficient and inexpensive if used with smart sprayers to precisely target unwanted plants. Unlike previous technologies, CNNs can distinguish between different weeds.
âFuture work will involve testing the CNNs for use on a smart sprayer and developing an application to provide growers with field-specific information,â the document said.
“Using CNNs to improve agricultural efficiency will result in significant cost savings for wild blueberry growers.”
Millions of savings
Recent figures confirm the economic importance of the harvest. In 2016, there were over 86,000 ha of fields in production in North America, producing approximately 119 million kg of fruit. Wild blueberries contributed over $ 100 million to the Nova Scotia economy in 2017, including $ 65.9 million in exports.
More than 100 weed species, however, limit yields by taking nutrients from the soil of wild blueberry plants and surly harvesting equipment.
In 2019, Scott White, a weed specialist at Dalhousie, discovered that sheep sorrel and hair fescue were the first and fourth most common weeds in the wild blueberry fields of Nova Scotia. They are usually managed by applying a uniform application of liquid herbicide with a boom sprayer attached to a farm tractor.
VelparÂ® was used to manage a range of weeds in wild blueberry fields starting in 1982, but is no longer used for hair fescue as it has developed resistance through repeated use. Curb â¢ SC, which is now used to manage hair fescue, is more than twice as expensive to apply.
âSo widely adapted in the wild blueberry sector, it could mean millions of dollars in agrochemical savings every year. This research can also be expanded to benefit other important cropping systems in Nova Scotia and Canada, âsays Dr. Esau.
Going forward, Dr Esau says CNNs will be trained for additional target weeds and one will be selected for use on a smart sprayer to selectively apply herbicides to wild blueberry fields. The selected CNN will be used in an application accessible on smartphones to help producers in field management decisions.
âThis is the first time that deep learning CNNs have been incorporated for use in wild blueberries for weed detection,â he says.
Dr Travis Esau, center, with the research team.