About the Project
AIGODS: Artificial Intelligence Grapevines Overwatch and Detection System
Coordinated by NOVA IMS, the AIGODS project addresses a fundamental challenge in modern agriculture: improving precision, transparency, and efficiency in vineyard monitoring and yield estimation. Operating within the complex terrain of the Douro Demarcated Region (DDR)—the world’s oldest officially demarcated wine region—the project leverages cutting-edge remote sensing and artificial intelligence to advance smart viticulture.
AIGODS pioneers the integration of high-resolution satellite imagery and very high-resolution orthoimagery with advanced AI techniques, including deep learning and genetic programming, to automate the identification of vineyard parcels and predict grape yield at multiple scales. This capability directly supports the regulatory mission of the Instituto dos Vinhos do Douro e Porto (IVDP), which oversees production quotas, quality standards, and geographic indications for the region’s globally renowned wines.
Rooted in geospatial intelligence, deep neural networks, and environmental data modeling, the project contributes to enhanced regulatory oversight, improved planning and quota allocation, and better adaptation to climate variability.
Through six structured work packages—ranging from literature review and data processing to AI model development, benchmarking, and dissemination—AIGODS delivers scalable, validated tools for automated vineyard monitoring. These outputs will help IVDP maintain an up-to-date vineyard cadastre and deliver early yield forecasts aligned with climate-smart agriculture principles.
As part of its implementation, the project will:
- Collect and process diverse datasets (remote sensing, meteorological, edaphological, phenological) covering 43,000 ha of vineyards across 169 administrative units;
- Develop deep learning models using CNNs, RNNs, transformers, and genetic programming to detect vineyard parcels and forecast yields;
- Benchmark multiple model architectures using performance metrics such as RMSE, IoU, and F1-score;
- Integrate predictive tools into IVDP’s regulatory workflows;
- Disseminate results through scientific publications, conferences, a public seminar, and media outreach.
