Detecting Vineyard Disease from Space Using AI Satellite Imagery with Deep Planet
- Growing Kent & Medway
- Jan 9
- 4 min read
Updated: 5 days ago
Deep Planet, in collaboration with Niab and leading UK vineyards, utilised our support to develop an AI-driven disease monitoring system. By analysing satellite imagery, the project successfully demonstrated a cost-effective solution to predict fungal infections, protecting the UK's growing wine sector from yield losses.
At A Glance: Project Quick Facts
Project Lead: Deep Planet (Sushma Shankar)
Collaborators: Niab (Dr Belinda Kemp), Gusbourne Wine Estates, Chapel Down Vineyards, Nyetimber, Rathfinny Wine Estates
Total Funding: £187,865 (Grant Awarded: £144,399 / Co-investment: £43,466)
Key Finding: AI models achieved over 90% accuracy in detecting Downy Mildew, Powdery Mildew, and Botrytis using high-resolution satellite imagery, offering a potential 40% increase in marketable wine yields.
Food System Areas:
The Challenge: Protecting a Growing Industry
The UK wine industry has experienced massive growth, quadrupling in size over the last decade. However, this success faces a rising threat from climate change, which creates favourable conditions for devastating fungal diseases like downy mildew, powdery mildew, and botrytis. These pathogens can severely impact crop quality and yield.
Traditionally, monitoring these diseases requires labour-intensive "scouting" by vineyard technicians or expensive drone surveys, costing growers thousands of pounds annually per hectare. As Sushma Shankar, co-founder of Deep Planet, explains, "While climate change has impacted the industry positively to grow better wines here, it has also started impacting it negatively in terms of the impact of disease on the crop."
The Innovative Idea: Precision Viticulture from Orbit
Deep Planet proposed a scalable, cost-effective solution: using satellite imagery combined with machine learning to identify disease "signatures" from space. By detecting the unique changes in leaf colour and other measurements caused by infection, they aimed to alert growers to outbreaks before they spread uncontrollably.
The core innovation lies in the accessibility of the technology. By integrating this disease prediction capability into their existing "VineSignal" platform, the project aimed to replace expensive manual checks with a digital service costing a fraction of the price – potentially saving growers thousands while safeguarding their harvest.
The Approach: Ground-Truthing the Data
Supported by one of our large R&D grant programmes, Deep Planet formed a powerful consortium with Niab and four of the UK’s most prestigious wine producers. The project relied on rigorous data collection to train the AI models.
Niab led the essential "ground-truthing" work, physically inspecting vines to verify disease presence and severity. Dr Belinda Kemp from Niab noted, "Our part of the project is doing the ground-truthing of all vineyards that are involved, and also juice analysis at harvest to have a look at the chemistry between grapes with and without disease." This ground data was then correlated with satellite imagery to teach the machine learning algorithms to recognise specific disease patterns.
The Results: Over 90% Detection Accuracy
The project successfully validated the technology, delivering impressive results. Using high-resolution satellite data, the AI models achieved detection accuracies of 91% for Downy Mildew, 94% for Powdery Mildew, and 92% for Botrytis.
These results demonstrate that satellite monitoring is a viable alternative to traditional methods. The system not only identifies current infections but helps predict future occurrences based on weather and soil conditions. Commercially, this tool empowers growers to target their interventions more precisely, reducing chemical usage and potentially increasing marketable yields by up to 40%.
See it Differently: Can Satellites Detect Vineyard Diseases?
Looking Forward: From Vineyards to Soil Carbon
With the technology proven, Deep Planet is moving rapidly to commercialise this feature for the UK market. The collaboration has ensured the product is practical for farmers, addressing real-world needs rather than just theoretical capabilities.
Following the success of this project, Deep Planet has secured further prestigious backing. They were recently selected as part of the first cohort of the ESA Phi-Lab UK, receiving nearly £190,000 in seed funding for their new 'SoilCarbonAI' project. This initiative will use geospatial AI to predict soil organic carbon levels, driving sustainable land management across regions lacking ground data.
Commenting on this new award, Sushma Shankar said: "This is a fantastic recognition of Deep Planet’s scientific approach to leveraging geospatial AI for critical environmental monitoring... accelerating the impact of space technology for a sustainable future."
Our Support: Accelerating Market Entry
This project was supported by one of our large R&D grant programmes, which was critical in bridging the gap between academic research and commercial application. The funding allowed Deep Planet to build the necessary prototype and access a network of major growers that would otherwise have been out of reach.
Reflecting on the impact of this support, Sushma said: "The funding has been transformational for us, and has enabled us to get the technology to market in less than a year... in other cases, we probably would have ended up four or five years down the line when, yes we would have had the technology, but then the impact would have been significantly less."
