Researchers from the University of Illinois used artificial intelligence (AI) to efficiently and accurately process precision agriculture data. The study addressed an increasing need in precision agriculture for data analysis solutions that can help farmers make management decisions.
Deep learning was used to generate yield predictions by incorporating information from topographic variables, soil electroconductivity, nitrogen and seed rate treatments. The new algorithm can detect site-specific responses to different inputs.
New research from the University of Illinois demonstrates the promise of convolutional neural network algorithm for crop yield prediction. Source: L. Brian Stauffer, University of Illinois
The team used data from a 2017-2018 Data-Intensive Farm Management project. During this project, seeds and nitrogen fertilizer were applied at varying rates across 226 fields in the U.S. Midwest, Brazil, Argentina and South Africa. On-ground measurements were paired with high-resolution satellite images from PlantLab to predict crop yield. The fields were digitally broken down into five-meter squares. Data on the soil, elevation, nitrogen application rate and seed rate were fed into a computer for each square. The goal was to learn how these factors interact to predict the yield in that square.
Using AI to untangle precision agriculture is a new concept. The team’s goal was to change how agronomic research is run by involving farmers directly using their machinery and fields. This experiment is likely just the beginning of the combination of precision agriculture and AI.
The study was published in Computers and Electronics in Agriculture.
