Supporting Beet Cultivation Research with AI and Flight Technology for Disease (SBR) Management
Main Objective: Develop and offer an AI-based digital phenotyping workflow using low-cost drones for accurate, large-scale SBR assessment.
Design Objective Ensure a user-oriented design of the AI-based digital phenotyping workflow.
- Craft a detailed protocol that defines the AI model’s requirements from user perspective, e.g., (i) which are the essential measures/features from the drone images for the SBR assessment (e.g. leaf area of yellow chlorosis or affected-healthy-plants ration, or similar). (ii) what is the minimum required accuracy for these parameters? We aim for at least an 80% stakeholder agreement on the AI-Model KPIs to ensure acceptance of the approach upon successful validation.
Development Objective: Develop and validate the AI-Model considering the AI-Model KPIs
- Support the 2024 field trial evaluation by providing a novel data source to the stakeholders.
- Collect data for the AI-model development and validation, thus, we aim to collect image data with low-cost drones at different field trial locations to ensure: (i) enough data amount, (ii) diversity in fields and image data, and (iii) recorded edge cases.
- Verify practical means of the low-cost drones & AI-model approach through a comprehensive comparison with the current state-of-the-art, i.e., the measures taken by the stakeholders by manual phenotyping in the field.
Market Objective: Prepare the launch of the drone based SBR phenotyping service in 2025 for both field trials and crop production.
- Formulate technical USPs of AI-based SBR assessment using low-cost drones and Pheno-Inspects web-based processing service.
- Create a business plan for this use case and
- Formulate an offer for the 2025 SBR field trials and optionally for the crop production domain.
Challenges and how they will be addressed
The ‘Syndrome of basses richesses’ (SBR) poses a considerable challenge in Central European sugar beet fields, resulting in economic losses estimated at around €128 million due to reduced sugar yield by up to 5%. Characterized by chlorotic leaves, necrotic taproots, and stunted growth, SBR complicates the harvesting process due to the altered rubbery texture of leaves which damages machinery. Spread primarily by cixiidae planthoppers and exacerbated by biotic stress factors, SBR has been expanding geographically since first being identified in Burgundy, France, in 1991.
Addressing this issue, sugar beet breeders, in collaboration with research institutions, are focusing on identifying tolerant varieties and effective agronomic practices. The introduction of BeetCraft-AID involves utilizing drone technology and AI-enhanced phenotyping analysis to monitor SBR more efficiently. This technology aims to identify characteristics during the growing season that indicate tolerance or susceptibility to SBR, facilitating proactive and strategic responses. This approach not only aims to improve crop resilience and yield but also enhances the precision of diagnosing SBR symptoms, surpassing traditional manual methods.
Tech components and data
- Low-Cost RGB Drones such as DJI Mavic3E
- Computer Vision & AI for Digital Phenotyping and Crop Monitoring
- High-Resolution RGB Data from Field Trials and Sugar Beet Fields
Expected outcomes
1
Design Phase Results: A protocol containing definition of target Image features, thus, KPIs for AI model performance as well as KPIs for an agronomic verification, i.e. a targeted correlation with the relevant parameters such as sugar content or yield loss. Protocol will be attached to the design phase report.
2
Development Phase Results: Collected datasets from different locations. AI model that can be deployed within our web-service and provides the target output. A report of the internal AI model validation showing an F1-score for the agreed phenotypic image features (AI model output) of at least 75%. Finally, an agronomic verification of the AI model results, showing a correlation of more than 80% between the results of the AI model and the actual objective of the phenotyping, i.e. the detection of the SBR infection, and its effect on sugar content and yield.
3
Market Phase Results: A Business plan to market further the solution to SBR assessment. Marketing Material in form of a success story and informative webinar, that demonstrates the technical USPs of AI-based SBR assessment using low-cost drones and Pheno-Inspects web-based processing service.
About Agrobit
Stay Connected
We appreciate your feedback.
Join our newsletter
Views and opinions expressed are however those of the author(s) only and do not necessarily reflect those of the European Union or Research Executive Agency. Neither the European Union nor the granting authority can be held responsible for them.