The Champagne wine region has taken a major step forward in vineyard health management by announcing a breakthrough in the detection of flavescence dorée using artificial intelligence.
The development was unveiled on January 29 during a conference on artificial intelligence in viticulture organized by InterLoire in Saumur, highlighting the growing role of advanced technology in safeguarding vineyards.
Flavescence dorée is one of the most serious diseases affecting grapevines, and its control remains a top priority for Champagne. The disease spreads rapidly, threatens vine longevity, and requires early and accurate detection to prevent widespread damage. Until now, monitoring has relied almost exclusively on human inspection, a method that is both labor-intensive and imperfect.
During the most recent survey campaign, 22,350 hectares of vineyards were inspected across the Champagne appellation. This effort mobilized 480 winegrowers and 80 volunteer experts, totaling 264 half-days of fieldwork, following 27,000 invitations sent to producers. Despite this large-scale operation, it was impossible to cover the entire 35,000-hectare appellation.
According to Mathieu Liébart, project manager at the Champagne Committee, the limitations of human inspection are significant. Detection accuracy averages around 50%, meaning that infected vines may be missed while healthy vines can be mistakenly flagged. This uncertainty increases both sanitary risk and operational costs.
To address these challenges, the Champagne Committee has been working with artificial intelligence for the past five years. The latest laboratory results are promising, with recognition rates exceeding 80% and reaching up to 94% in some cases, particularly for Chardonnay leaves collected in July. These results were achieved by training AI models on extensive image datasets that include healthy leaves and leaves affected by flavescence dorée as well as other diseases that may cause visual confusion.
For now, these performances remain confined to laboratory conditions. Applying the technology directly in vineyards—using cameras mounted on tractors or drones—would require processing massive volumes of data in real time, a technical hurdle that has yet to be overcome. Nonetheless, research continues, and the Champagne Committee remains confident that field deployment will eventually become feasible.
This breakthrough signals a shift toward more precise, data-driven vineyard monitoring, offering Champagne producers a powerful new tool in their fight against one of viticulture’s most persistent threats.
Source: Vinetur