A New Era of Data-Driven Decision Making
Digital phenotyping is emerging as a transformative tool in agriculture and horticulture. By collecting and analysing observable traits through digital imagery and synthetic datasets, this approach enables more accurate and informed decisions across breeding and crop management programmes.
Complementary Expertise: AbacusBio and Synetic AI
AbacusBio sees strong potential in combining its genetics expertise with the artificial intelligence capabilities of Synetic AI, a branch of US-based consulting firm The Main Branch, to deliver innovative solutions for breeding programmes.
“We want to use this digital phenotyping data to enhance and enrich breeding programmes, so we can have more data with higher levels of accuracy going into the selection process,” says Martin Howes, consultant at AbacusBio Canada.
How Digital Phenotyping Works
David Scott, founder and CEO of The Main Branch and Synetic, explains that creating their digital representation of a cattle beast begins with identifying skeletal structures such as eye distance, bone width, femur and patella size. From there, 524 key points are mapped onto the animal, feeding into an AI model that identifies unique traits and correlations.
“Instead of waiting for the world to give us data, we create it synthetically based on expert knowledge,” says Scott. Millions of images are generated, and computer models are trained to make predictions and inferences.
Applications in Breeding and Yield Forecasting
Howes sees exciting potential in using AI to analyse physical or behavioural characteristics, which could lead to tools that improve breeding outcomes and retail beef yield, for example. Scott notes that selecting animals with favourable traits for specific environments could be highly beneficial. For crop farmers, forecasting yield is another promising application. AI can begin tracking plants from the moment they emerge from the soil – monitoring growth rate, stem count, and other indicators. These insights feed into predictive models that estimate yield.
“We’re asking ourselves how many interesting things we can do,” said Scott.
From Data to Value
Howes is particularly interested in the genetic insights that can be extracted, with AI-powered video camera models integrated into equipment like sprayers and harvesters.
However, he also highlights a challenge: bridging the gap between data collection and value creation. AI enables detailed analysis of traits like drought resistance but translating that into a breeding value in a genetic improvement programme takes specialist skills and a lot of work.
“AbacusBio are experts in bridging the gap between large phenotype data sets and creating breeding values that breeders can use to select improved plants and animals for the future needs of the planet,” says Howes.
Ensuring Accuracy and Consistency
John Crowley, Managing Director of AbacusBio Canada, raises a key point about data quality: “How does accuracy for one trait hold up if the data collected varies in quality? And how does that affect breeding outcomes?” Scott emphasized the importance of consistency and accuracy in rendered datasets. As AI becomes more embedded in the breed selection process, the data will grow in depth and reliability – eventually reducing human bias and error.
A Glimpse into the Future
Imagine a large-scale feedlot where a camera mounted on a pole collects data on hundreds of steers in minutes. An AI model could be used to monitors pens of animals for slaughter weight, health, and growth rates. As animals pass through gates for food and water, AI-powered systems could capture and analyse traits – reducing the time-consuming process of manual weighing and checking.
Data will be live, real-time, and frequent. Growth patterns will be linked to genetic tags and predictions, enabling more precise management.
“This is still in the future, but it is the kind of capability this technology is unlocking,” says Howes.
Crowley agreed the advantage of AI is that you can collect and analyse many, many more animals or plants, providing the AI gives you an accurate enough measure. It can greatly increase the scale of what you can measure.
Crowley adds, “And you can’t improve what you can’t measure.”
Preparing for the ‘New Norm’
Howes believes AbacusBio must stay ahead of this wave, calling it the ‘new norm’ for integrating data into breeding programmes. As costs decrease, phenotyping will become more economical and influence broader management practices. AbacusBio is particularly focused on applying this technology at the industry level so all farmers can benefit.
“ The expansion of AI’s capabilities means the ability to capture huge amounts of performance data, which will result in greater rates of genetic improvement in our plants and animals. There’s change coming. The “when” and “how” of that change is still unfolding.” Martin Howes
