Innovative Fertilizer Identification through Artificial Intelligence

Finding the right fertilizer in very large spreading charts is a challenge for people working in agriculture. The process is time-consuming and prone to errors. Incorrect settings can not only reduce yields but also cause unnecessary environmental impacts. Accordingly, there is a high demand for a reliable, practical solution.

Fertilizer classification using AI

In a joint project, DFKI and AMAZONE combined their respective strengths in a relationship of trust and close partnership. Building on AMAZONE’s extensive database containing image material and documented spreading characteristics, the project partners developed innovative AI methods for automated fertilizer classification.

The shared goal was to create an AI system that provides tangible relief for agricultural professionals: a solution that reliably identifies the fertilizer used based on a simple photo or—if the fertilizer is not yet known—determines a fertilizer with comparable spreading characteristics. Research and industry worked closely together, with a clear focus on practical applicability and user benefit.

AI that builds trust—and simplifies processes

The developed artificial intelligence analyzes the photo of the fertilizer and generates a digital fingerprint with more than 250 physical parameters. This fingerprint is automatically compared with the datasets in the existing fertilizer database. Within seconds, the most similar fertilizer can be identified—and basic settings for the fertilizer spreader can be derived immediately.

From research to application: synergies in technology Transfer

The results of the project provided the foundation for the development of the EasyMatch app. The AI-based recognition system enables fertilizers to be captured using a smartphone camera in combination with a handy reference frame. From the photo, a digital fingerprint is generated, which the AI compares offline with the fingerprints stored in the fertilizer database.

With just the press of a button, agricultural professionals receive precise setting

 recommendations for unknown fertilizers—independent of product name, manufacturer, or packaging. For the first time, spreaders can also be reliably set for previously unknown spreading materials. This saves up to 30 minutes per spreading start, significantly reduces incorrect settings, and sustainably improves work quality.