AI4SusCo - Artificial-intelligence-based classification system for deforestation-free coffee
The AI4SusCo project tackles the pressing issue of deforestation associated with coffee imports. In response to the European Union's EUDR Regulation, which bans imports of agricultural products linked to deforestation after December 31, 2020, the project aims to develop advanced technological solutions by employing deep learning methods for accurate detection of coffee crops and forests.
General Project Objectives
The AI4SusCo project plans to apply, evaluate and optimize deep learning methods for forest and coffee crop detection, such as Convolutional Neural Networks (CNNs) and newer spatio-temporal methods. In this context, several studies show the added value of derived and modelled data, such as vegetation indices, hydrological and meteorological data. This data will be successively integrated into the models to be developed in order to determine deforestation as accurately as possible.
The project aims to develop a technological solution that makes coffee certification more efficient and accurate, thereby making an important contribution to the sustainable development of the coffee industry, the preservation of rainforests and saving our global climate. In addition, the new EU regulation means that there is an immensely growing sales market for this type of certification
At a Glance
| Category | Description |
|---|---|
| Research project | AI4SusCo - Development of an artificial intelligence-based classification system for deforestation-free coffee for certification under the EU sustainability regulation. |
| Administration | Project Lead: Prof. Dr. Lars Ribbe Personal Profile |
| Faculty | Faculty of Spatial Development and Infrastructure Systems More |
| Institute | Institute for Technology and Resources Management in the Tropics and Subtropics (ITT) More |
| Persons involved |
M.Sc. Juan Ramirez Student M.Sc. Juan Mercado |
| Partners |
- GRAS - Global Risk Assessment Services GmbH: independent environmental service provider that offers information and advice on environmental and social sustainability. - Institute for Information Science at TH Köln (IWS) |
| Sponsors |
Federal Ministry for Economic Affairs and Climate Action (BMWK) Zentrales Innovationsprogramm Mittelstand |
| Duration | June 2024 - May 2026 |
Specific Project Objective(s)
- Developing a comprehensive concept document that outlines the requirements and functionalities needed for R&D and prototype development.
- Completing the acquisition of satellite data from Sentinel-1 and Sentinel-2 for Colombia and Brazil, including the labeling of derived and modeled data.
- Performing data preprocessing on satellite data and assessing the data quality and its predictive power using Multicriteria Analysis (MCA) to ensure it is ready for analysis.
- Developing and optimizing machine learning models using the preprocessed derived and modeled data to ensure accurate predictions.
- Global Transfer: Evaluating the applicability of the developed models in transfer countries and implement necessary adjustments based on performance feedback.
- Evaluation: Conducting a comprehensive evaluation of the mapping results and model performance through on-site assessments and user feedback.
- Prototype: Creating a fully functional prototype that integrates the front end, backend, API, and is deployed as a Docker container for scalability.
Funding Reference
Logo BMWK/ZIM (Image: BMWK/ZIM)