See Disease Before It Strikes: AI-Powered Protection for Vineyards

How Deep Planet Predicts Vineyard Disease with AI and Satellites 

In viticulture, disease disrupts entire growing seasons. Early detection is key to protecting yields, maintaining grape quality, and avoiding unnecessary input costs. 

But traditional methods for identifying disease in vineyards are reactive, slow, and expensive. They rely on manual sampling; agronomists walking rows and inspecting the canopy, ultimately extrapolating risk from limited physical data. It’s time-consuming, labour-intensive, and often too late. 

At Deep Planet, we’re taking a proactive, data-driven approach that uses artificial intelligence, satellite imagery, and ground truth data to forecast disease probability and reduce detection times compared to traditional methods. 

The Problem: Disease Doesn’t Wait 

Growers are especially vulnerable when: 

  • Recent weather events raise the risk of infection 

  • In-field early disease detection requires significant expertise 

  • Resources are limited and action must be precise 

The Solution: AI-Powered Disease Probability Maps 

Deep Planet’s disease prediction tool provides: 

  • Per-pixel disease risk maps across entire vineyard blocks 

  • Updated predictions available at near daily frequency 

  • Accurate detection of the 3 most prevalent diseases: 

  • Downy Mildew (Plasmopara viticola) 

  • Powdery Mildew (Uncinula necator) 

  • Botrytis Cinerea 

  • AI models trained using: 

  • Sentinel-2 (10m) and PlanetScope (3m) imagery 

  • Vegetation indices (e.g., NDVI, GNDVI, SIPI) 

  • Weather, soil, and terrain data 

  • Ground-truth observations collected by agronomists 

  • Near-90% detection accuracy in field trials 
    (89.6% DM, 93.7% PM, 91.5% Botrytis) 

These maps highlight where it’s likely to emerge, helping growers focus scouting and treatment directly where needed. 

How It Works 

Using our AI models, we simulate a predictive disease scenario that shows spatial variation in risk across your vineyard: 

  1. Environmental triggers (rain, humidity, heat) increase the probability of infection 

  1. Spectral imagery identifies early stress signals in vegetation 

  1. Our ML pipeline combines this data with grower-specific factors to produce disease likelihood maps 

  1. Growers act: Target sampling, apply treatments, and avoid blanket spraying 

Why This Matters for Growers 

  • Avoid over-spraying by targeting only the blocks at risk 

  • Improve scouting efficiency and reduce labour 

  • Save time by catching early-stage infections before visual symptoms are identified 

  • Use your own field data to refine the model further 

And most importantly: this tool empowers growers to be in control, rather than reacting after the damage is done. 

Results You Can Trust 

In field trials across four UK vineyards (Chardonnay & Pinot Noir), our AI models were validated using European disease protocols and ground-truth data collected from August to October 2024. 

  • Models accurately flagged early-stage infection zones 

  • Growers used maps to refine spraying and sampling 

  • Risk maps were updated regularly to account for changing conditions 

“Supported by our Prototyping and Demonstrator Fund and access to the Wine Innovation Centre, Deep Planet has created a precision viticulture solution that helps growers identify disease patterns, optimise harvesting, and improve crop quality—bringing their technology to market in under a year.” 
Dr. Belinda Kemp, Group Leader for Viticulture and Oenology Research, NIAB East Malling 

Our real-time insight allows growers to adapt with precision whether they’re managing outbreaks triggered by weather or fine-tuning treatment at the block level. 

Ready to Predict Before You React? 

Don’t wait for disease to appear. Let Deep Planet help you spot it early—with confidence and precision. 

Book a demo today to see your vineyard through the lens of AI: https://calendly.com/hello-deepplanet/30min?month=2025-09  

 

Next
Next

Deep Planet and National Trust - Habitat and Biodiversity Mapping for Conservation