The use of artificial intelligence in wine is no longer limited to experimental research—it is rapidly expanding across viticulture and oenology, from disease detection in vineyards to sensory analysis of finished wines.
Recent presentations at the InterLoire conference in Saumur illustrated how deeply AI is beginning to influence the wine sector.
Beyond Champagne’s work on flavescence dorée, the Pl@ntNet consortium is advancing AI-driven plant and disease recognition through a participatory approach. Known for its widely used mobile application with over 25 million users, Pl@ntNet can already identify numerous plant species from photographs. Disease recognition, however, presents a more complex challenge.
Currently, Pl@ntNet can identify seven foliar diseases in rapeseed and is expanding its scope to include varietal identification within species and biodiversity assessment. Lydia Bousset-Vaslin, researcher at INRAE in Rennes, emphasized the importance of collaboration, inviting the wine sector to contribute annotated images of grapevine diseases. This citizen-science model, which played a key role in the app’s success, remains central to its future development.
To analyze images, researchers rely on advanced models such as DINOv2, developed by Meta. These systems require vast and diverse image databases that capture all stages of disease development across multiple grape varieties. The more comprehensive the dataset, the more reliable the algorithm becomes.
Artificial intelligence is also transforming wine analysis beyond the vineyard. Stéphanie Marchand-Marion, professor at the Bordeaux Institute of Vine and Wine Sciences, presented results from studies using AI to interpret wine chromatograms. In 2023, a joint French-Swiss project demonstrated that AI could distinguish wines from Saint-Émilion and Médoc when traditional analytical methods failed.
Further research by a Spanish team showed that AI could describe the taste characteristics of 30 Spanish and Australian wines almost as accurately as a professional tasting panel. By combining chromatography and voltammetry with machine learning, researchers achieved faster and more cost-effective sensory evaluations than conventional human assessments.
Marchand-Marion stressed that the wine sector must take ownership of these technologies and actively communicate about their use. Without engagement from producers and institutions, there is a risk that external actors may interpret wine data inaccurately or without sufficient context.
While practical, large-scale application in vineyards and wineries still faces technical constraints, the progress achieved so far points clearly toward deeper integration of artificial intelligence in wine. From vine health to market positioning, AI is poised to become a strategic tool shaping the future of the industry.
Source: Vinetur
0 comments