Stellenbosch University researchers are harnessing artificial intelligence to streamline the identification of grapevine leafroll disease, a significant threat to South Africa’s wine industry.
A groundbreaking research project led by Prof Lizel Mostert at the Department of Plant Pathology, Stellenbosch University, is set to change how vineyard managers and workers identify grapevine leafroll. This disease poses a major challenge to grape production in South Africa. Funded by South Africa Wine, the project, titled “Automated Visual Assessment of Grapevine Leafroll Symptoms Using Artificial Intelligence,” began in 2023 and promises to significantly enhance disease identification with its innovative approach.
Â
Cutting-edge technology for precision agriculture
The project aims to develop an artificial intelligence (AI)-based system that can automatically recognise grapevine leafroll-associated virus type 3 (GLRaV-3) symptoms in red and white wine cultivars using smartphone photos. This system will drastically reduce the time and effort required to identify infected vines. This user-friendly, automated tool will benefit vineyard managers and workers, who often rely on extensive online resources or physical manuals to diagnose diseases.
“Automated recognition through smartphone images will not only save time but will also enhance accuracy in identifying symptomatic vines,” said Anel Andrag, the Research, Development and Innovation Manager (Viticulture) at South Africa Wine. “This technology will enable more efficient management of grapevine health, ensuring better outcomes for producers and the wine industry as a whole.”
Â
Addressing a major agricultural challenge
Grapevine leafroll is an economically significant disease that affects grapevine production across South Africa. Effective identification of vines infected with GLRaV-3 is crucial for maintaining clean propagation material and managing infected vines, particularly in newly planted vineyards and mother block vineyards. The disease’s symptoms can be easily confused with those of other conditions, such as potassium deficiency, Shiraz disease, and Aster yellows, making accurate diagnosis challenging.
Currently, no disease identification app is specifically designed to detect grapevine leafroll. The project led by Prof Mostert seeks to fill this gap, providing a vital tool for growers to distinguish leafroll from other diseases and nutritional disorders more accurately.
A game changer for the wine industry
The outcomes of this research will have significant implications for vineyard and nursery management. By enabling more precise identification of leafroll-infected vines, the AI-based system will support better decision-making and disease management practices. As the project progresses, the team at Stellenbosch University is optimistic about its potential impact. “Our goal is to provide a practical solution that directly addresses the needs of the industry,” Prof Mostert emphasised. “With the integration of artificial intelligence, we can offer a powerful tool to help growers manage their vineyards more effectively and sustainably.” The current project includes seven red and three white cultivars for which a validated photo database of healthy and leafroll-infected leaf symptoms of approximately 27Â 000 photos have been generated. The power of using neural networks (a specific application in AI) is achieved by having an image database including all possible variations of symptoms.
With its innovative approach and potential to aid in leafroll identification, this project represents a significant step forward in the fight against grapevine leafroll disease. Developing a smartphone app will be the next step in the process, as well as expanding the image database with additional cultivars and ongoing validation of newly submitted images. The use of AI in agriculture continues to grow, and projects like this one at Stellenbosch University are at the forefront of this technological evolution, offering hope for a healthier, more productive future for the wine industry.
Image: Freepik AI