Overview
EdenCore has partnered with agrifoodTEF France and Institut Francais de la Vigne et du Vin (IFV) to bring AI-driven early yield estimation to vineyards, combining cutting-edge agronomy, applied AI, and decades of viticultural expertise. Together, they are validating a farmer-centric technology that delivers real-time, reliable insights and bridges the gap between research, innovation, and everyday vineyard operations.
France
- Tree Crops
- Viticulture
The challenge
Winegrowers today operate at a crossroads of increasing pressures: fluctuating climates, soaring production costs, labour shortages, and the need to “grow more with less,” as EdenCore’s Head of Agronomy, Dimitra Dafnaki, explains.
“Fields are rarely uniform,” she notes. “Managing different zones effectively is a big challenge, and this is where AI and precision agriculture tools can make a real difference.”
Farmers also face labour-intensive tasks such as scouting and sampling, activities that, despite being essential, demand time, specialised knowledge, and consistency. Automation can lighten this load by generating real-time field insights that integrate seamlessly into existing workflows.
For IFV, early and accurate yield assessment is particularly valuable. As engineer Christophe Gaviglio highlights, “anticipation of the harvest and preparation of the wineries” rely on reliable estimates, something vineyards have historically lacked.
When EdenCore approached IFV, the institute immediately recognised the relevance of the use case. They offered access to an experimental vineyard containing two grape varieties (one red and one white) each managed with both defoliated and non-defoliated treatments, creating an ideal environment for testing and validating AI models.
The agrifoodTEF service
At the core of this collaboration is Viewer, EdenCore’s tractor-mounted, AI-powered camera system. Designed as a highly adaptable field companion, it delivers three main functionalities:
- Yield monitoring from bloom to harvest: The capability assessed within the agrifoodTEF project. Viewer performs continuous scouting during the season, capturing visual data to predict yields and track growth stages with increasing accuracy.
- Real-time variable rate application (VRA) spraying: Using canopy density analysis, Viewer enables precise spraying, reducing input waste and improving efficiency.
- Vigor and disease mapping: After each field pass, the system automatically generates vigor maps, yield maps, anomaly detections, and field-level analytics accessible on EdenCore’s online dashboard.
Much of the development depends on the ground-truth measurements collected in the field, both to calibrate the system across grape varieties and to validate its precision. “Data is integrated into the algorithms, allowing the system to provide more accurate estimations,” Dafnaki explains.
However, vineyard environments introduce technical challenges, especially occlusion, when foliage, branches or other fruits obscure the system’s view. EdenCore addressed this through two complementary strategies: equipping Viewer with LED lights on both sides to reduce shadow-induced occlusions, and continuously collecting ground-truth samples that help the algorithm account for hidden fruits.
The system is designed with flexibility in mind. Its adjustable mounting options allow it to adapt to different crop heights, vineyard structures, and machinery, an essential feature for practical, everyday use.
“The feedback from farmers has been very positive,” says Dafnaki. “They appreciate that the system integrates into their daily operations without disrupting their routines.”
Explore the AI Dataset Service
The impact
The collaboration between EdenCore and AgrifoodTEF’s French node represents more than a technology trial: it is a shared effort to accelerate the digital transformation of viticulture through field-validated, farmer-driven innovation.
For EdenCore, the partnership has expanded their dataset with high-quality measurements collected under carefully controlled conditions, an invaluable asset for improving prediction models. “The product is not definitive yet” Gaviglio remarks, “and that’s precisely why the collaboration is useful. It helps EdenCore enrich its database and refine the technology.”
For the wider sector, early results show a clear trajectory: scalable, adaptable solutions capable of extending beyond vineyards to crops like blueberries, pistachios, and almonds, where EdenCore is already expanding its activities.
Looking ahead, both partners see AI becoming increasingly influential in viticulture, as long as the use cases remain relevant and the return on investment is clear. Yield prediction stands out as one of these high-value fields, with strong potential for broader market adoption in the next decade.
Ultimately, collaborations like this will play a decisive role in shaping the future of sustainable agriculture. As Dafnaki concludes, “They bring together farmers, technology providers, and research institutions, stakeholders who each hold part of the puzzle.”
And, in Gaviglio’s words, they empower small and innovative companies by giving them access to experimental facilities and ground-truth data, resources essential to bringing trustworthy, scalable solutions to market.
Through such partnerships, agricultural innovation doesn’t just advance, it takes root.
