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: Development of an AI-based expert system to assess the effectiveness of weed control measures under consideration of the damage threshold principle using the example of sensor-guided hoe - subproject A (WeedAI)
Project
Project code: 28DK121A20
Contract period: 19.04.2021
- 18.10.2024
Budget: 498,053 Euro
Purpose of research: Experimental development
Keywords: crop production, crop protection, plant health, data collection, integrated plant protection, precision farming, sustainability, weed, sugar beet, AI Artificial Intelligence, agricultural engineering
Methods of plant recognition (Deep Learning, AI) derived from basic scientific research should make progress for applications in agriculture. Vision-processing methods will be used to differentiate between crop and weed plants as well as between weed species. Taking into account the damage threshold principle, the potential damaging effect or control worthiness can be estimated, the work quality / weeding success of (sensor-guided) hoes in the crop (row recognition and control) and the effectiveness of their tools can be evaluated. Sensor-guided hoeing machines should relieve drivers, increase area output and also control weeds more effectively. Accordingly, new developments in industry and research are a highly topical issue. From a scientific point of view, not much is known about the actual benefits, but they are important for farmers, contractors and machinery rings in their purchasing decisions. In order to close this knowledge gap, a reproducible test method that is as objective as possible is to be developed in conjunction with the vision-control methods described above. Users of machine hoes and manufacturers get the opportunity to test and further develop their technologies (finished products or prototypes). A facilitated automated assessment of weed control strategies also supports the testing process considerably. This results in three goals: 1. further development of existing deep learning methods and algorithms for the assessment of existing crops, weed flora and their development after crop protection measures. 2. development of a reproducible test procedure of alternative plant protection methods using the above described plant recognition. 3. application of plant recognition and test methodology for the assessment of weed status and sensor-guided hoes.
Section overview
Subjects
- Crop Production
- Crop Protection
