We use cookies on our website. Some are necessary for the operation of the website. You can also allow cookies for statistical purposes. You can adjust the data protection settings or agree to all cookies directly.
Data Protection Declaration
Imprint
Collaborative project: AI-Based valuation system for plant level autonomous valuation - subproject B (BoniKI)
Project
Project code: 28DK120B20
Contract period: 01.05.2021
- 30.04.2024
Budget: 244,388 Euro
Purpose of research: Applied research
Keywords: crop production, digital world, maize, precision farming, wheat, weed, crop protection, plant health, AI Artificial Intelligence
The goal of BoniKI is the development of a multi-level AI approach (AI=artificial intelligence) for the automatic, plant-specific scoring of single plants and plant canopies in agriculture. By means of a comprehensive data acquisition with unmanned aerial systems (UAS) and the use of the latest AI methods for the recognition and evaluation of plant characteristics, the manual assessment, i.e. the expert evaluation and scoring of plant characteristics relevant to agriculture, will be supplemented by a robust, automated solution. In this project, agronomic experiments are evaluated by experts using classical manual assessment. At the same time, high-resolution image data (in the mm range) will be collected during assessment flights with a UAS system and evaluated automatically. In an instance segmentation network (e.g. MaskRCNN or ShapeMask) single plants shall be recognized robustly, even if the complex scenarios are difficult to recognize because e.g. leaves overlap or shade each other.Due to occlusions and the high natural variability in appearance, previous methods have not yet been able to solve these satisfactorily. Thus, BoniKI digitizes and automates the traditional scoring procedure, improving the traceability, objectivity of the observation and as a result, the quality of the outcome. Furthermore, it is made available for further applications and evaluation steps. Modern AI methods not only increase the robustness and adaptability of the recognition process, but also apply automated learning to the assessment process itself, which encodes and continuously expands the knowledge base and allows new insights into the relationship between optical characteristics and effects on plant growth.
Section overview
Subjects
- Crop Production
