Researchers at the University of Bonn have developed groundbreaking software that leverages artificial intelligence to simulate the growth of field crops. Through the use of thousands of photos from field experiments, a learning algorithm was trained to predict the future development of cultivated plants based on a single initial image. This innovation allows for accurate estimation of parameters such as leaf area and crop yield.
The innovative software opens up a realm of possibilities for farmers seeking to optimize their crop yield. By utilizing drone photos and the learning algorithm, it becomes possible to virtually simulate different scenarios. For example, farmers can now assess the impact of using manure instead of artificial fertilizers on crop development.
The software facilitates decision-making by providing insights into how interventions such as pesticide use or fertilization will affect crop yield. By feeding the program with drone photos from various growth stages, researchers were able to document the development of crops like cauliflower under specific conditions. This data was then used to train the learning algorithm, which could generate images depicting the future growth of crops based on an initial aerial image.
Moreover, the software is equipped with a second AI program capable of estimating key parameters like crop yield from plant photos. This feature enables early predictions of cauliflower head size, offering valuable information in the early stages of the growth period.
A key focus of the research is on the use of polycultures, where different plant species are grown together in a single field. The software can analyze various mixing experiments to determine the compatibility of different plants and their ideal ratios for optimal yield. Polycultures not only enhance yield but also make crops less vulnerable to pests and environmental factors.
While traditional models rely on a fundamental understanding of plant nutrients and environmental needs, the software developed by the University of Bonn is solely based on experience gained from training images. Moving forward, researchers aim to explore how the software can be combined with process-based models to improve forecast quality.
Overall, the software represents a major leap towards the intelligent digitalization of agriculture, paving the way for more sustainable and productive farming practices. The findings of this study, published in Plant Methods, highlight the potential of AI-driven technology to revolutionize crop management and decision-making in agriculture.