Physical services
Overview
Winegrowers across Europe are under growing pressure to do more with less: climate variability, rising costs, and labour shortages are reshaping how vineyards are managed. Within agrifoodTEF's French Node, agritech startup EdenCore partnered with the Institut Français de la Vigne et du Vin (IFV) to validate an AI-powered, tractor-mounted camera system capable of estimating grape yields early in the season, directly from the field.
The collaboration combines EdenCore's applied AI expertise with IFV's decades of viticultural knowledge, testing whether continuous, automated crop monitoring can give farmers reliable insights that have historically been difficult to obtain.
Impact
Testing within a real vineyard environment gave EdenCore access to high-quality, controlled data that would otherwise be hard to collect: an experimental plot with two grape varieties, each managed under both defoliated and non-defoliated conditions. This let the team validate the system's predictions against carefully documented ground-truth measurements, strengthening the underlying models.
For IFV, the pilot addressed a long-standing gap in the sector: the ability to anticipate harvest volumes early enough to inform winery preparation. Beyond the immediate use case, early results point to broader applicability, with EdenCore already exploring extensions of the technology to other tree crops such as blueberries, pistachios, and almonds. The project illustrates how agrifoodTEF helps connect emerging AI providers with research institutions and real field conditions, accelerating the path from prototype to farmer-ready tool.
The Challenge
Vineyard management is inherently uneven: fields vary from zone to zone, and the manual scouting and sampling needed to track crop development are time-consuming, require specialised expertise, and are hard to keep consistent at scale. Reliable early yield estimates, in particular, have long eluded winegrowers, making it difficult to plan for harvest logistics and cellar capacity in advance.
For EdenCore, the challenge was to build a system able to operate reliably under highly variable field conditions, while remaining simple enough to fit into farmers' existing routines without disrupting daily operations.
The Solution
Through agrifoodTEF, EdenCore tested Viewer, its tractor-mounted, AI-powered camera system, within IFV's experimental vineyard. The platform combines three capabilities: continuous yield monitoring from bloom to harvest, variable-rate spray application based on canopy density, and automated vigour and disease mapping accessible through an online dashboard. The yield-monitoring function was the specific capability assessed within the agrifoodTEF project.
To address occlusion, a key technical hurdle in vineyard imaging caused by foliage and fruit blocking the camera's view, EdenCore equipped Viewer with side-mounted LED lighting and continuously collected ground-truth samples to help the algorithm account for hidden fruit. The system's adjustable mounting also allows it to adapt to different crop heights and vineyard structures, supporting use across varied farm setups.
By pairing EdenCore's technology with IFV's field expertise and experimental infrastructure, the project moved the yield-estimation model closer to real-world reliability, while reinforcing agrifoodTEF's role in helping innovative SMEs validate their solutions under genuine operating conditions before wider market rollout.
