Adopting this new vineyard technology is critical to the California grape industry.
—Tom Shapland, PhD, CEO and cofounder of Tule Technologies
Anything humans can observe with their eyes a computer can be taught to see as well. Growers scout their vineyards for pests, disease, water stress, vigor, and more—soon this will all be achieved by computer vision: In the near future, vehicle-mounted camera systems will provide fast, efficient vineyard data and alert growers to problems that may require further human inspection and/or intervention.
But vineyard scouting is just one area of viticulture that Artificial Intelligence (AI) will transform.
Today, the term AI appears all over popular media. But in order to use the technology effectively within vineyard and winery operations, we have to learn what it is and how it works.
What is AI?
AI is a broad concept, describing a computer system that performs tasks that we would expect to require human intelligence. For example, a computer system like Amazon’s Alexa understands spoken language, the same way we expect people to understand spoken language. That is AI. A computer system that counts grape clusters utilizing the camera on a smart phone—that is AI. This second example is where we have seen the most striking gains in the power of this technology, the area of computer vision.
What is leading to these new capabilities in computer vision over the last few years is the use of machine learning. Put simply, creating software is like baking. You start with some inputs (i.e., flour, water, salt, and yeast) and use a recipe to produce an output (i.e., bread). In traditional software development, engineers write every step in the recipe. In machine learning, software engineers give a computer many examples of the inputs and outputs, and the computer learns from the examples to develop its own recipe. AI has become especially adept at looking at pictures as inputs and using a machine-learned recipe to tell us what is in the pictures—the output.
At Tule, our team of engineers have developed Tule Vision, a computer vision model that monitors water stress levels. Growers are able to take a short video of their plants, and the AI tells them the midday leaf water potential (i.e., LWP, or “thirstiness” of the plants). To achieve this, we gave a computer thousands of pictures of vines, as well as labels for each picture describing the midday LWP of the vines. The computer then learned to recognize the different levels of water stress in the plants, just as an experienced viticulturist learns by observing the vine canopy vigor, leaf-petiole angle, color, etc.
In the next few years, the wine industry will welcome many more AIs similar to Tule Vision that will provide agronomic insight via pictures.
I leave you with a warning: If California growers do not adopt AI, the grape industry will be wiped out. Though many growers chose their career path because they like working outside with their hands and not with computers, they need it will ultimately behoove them to seek out the latest automation tools and integrate them into their daily workflow. Doing so will increase the efficiency of daily and seasonal tasks, mitigate concerns over labor shortages, and allow our growers to make informed, real-time decisions about the health of their vines, grapes, and resulting wines.
If we want a vibrant grape and wine industry, and for future generations to have access to local wines and produce, then growers will have to eventually learn to farm by computer.
Get involved or get left behind.
Tom Shapland, PhD is the CEO and co-founder of Tule Technologies. He has a doctorate in horticulture and agronomy from UC Davis. You can learn more about how Tule helps growers make irrigation decisions at www.tuletechnologies.com.
Shapland will be speaking at WINExpo 2021. The AI and Automation in Vineyards panel discussion will feature vintners currently utilizing this advanced technology in the field who can speak, first hand, to their successes. Learn more about WINExpo and register.