Virtual services
Overview
QZ Solutions is a Polish technology company developing digital tools for precision agriculture. Its flagship system, SoilEO, combines hyperspectral imaging with AI algorithms to map the nutrient content of agricultural soils and help farmers decide how much fertiliser to apply, and where.
To make its models more accurate, QZ Solutions turned to the agrifoodTEF Polish Node, part of the European Testing and Experimentation Facility for AI and robotics in the agri-food sector. The support came from two partners. The first was the Łukasiewicz – Poznań Institute of Technology (Łukasiewicz – PIT), a research institute within Poland’s Łukasiewicz Research Network. The second was the Wielkopolska Agricultural Advisory Centre in Poznań (WODR), which advises farmers across the region. Together they gave the company access to benchmark datasets and enriched annotations to validate and optimise SoilEO’s algorithms, which brought the system closer to market.
Impact
Validation against high-quality reference data has made SoilEO more accurate and more reliable, so farmers receive more precise fertilisation recommendations. According to QZ Solutions, the enriched datasets have shortened the time needed to deliver advisory services and strengthened the system’s ability to detect both macro- and microelements across a range of soil types. The company considers SoilEO ready for deployment and expects at least 20 users by 2027.
The benefits extend beyond the farm gate. Applying fertiliser only where and in the quantities it is needed lowers input costs and reduces the environmental footprint of farming, in line with Europe’s goals for healthier soils and more sustainable food production.
The Challenge
Farmers are under pressure to keep yields high while cutting fertiliser use and nutrient losses. Those losses pollute water and degrade soils. European policy reflects this: the Farm to Fork Strategy aims to reduce nutrient losses by at least 50% and fertiliser use by at least 20% by 2030. The EU Mission A Soil Deal for Europe places soil health at the centre of the transition. Precise fertilisation depends on knowing what is actually in the soil, but conventional soil testing is expensive and slow, and it is hard to carry out at the scale of whole farms or regions.
SoilEO was designed to close that gap by producing AI-generated maps of key soil parameters, including magnesium, potassium, phosphorus and pH. An AI model is only as reliable as the data it is checked against, however. To validate and fine-tune its algorithms, QZ Solutions needed a large set of high-quality, geolocated soil samples with laboratory reference values, consolidated into datasets fit for benchmarking. Building such a dataset alone would have taken the company considerable time and money.
The Solution
Through Łukasiewicz – PIT and WODR, agrifoodTEF provided two complementary services. The first was the generation and provision of labelled datasets for benchmarking. More than 2,000 geotagged soil samples were collected in the field and analysed in the laboratory. The resulting reference dataset covers five single and 20 double attributes describing nutrient levels and their valuations. The second service was the curation and provision of annotated datasets. Here the samples were enriched with geotagged photographs, digital maps and structured annotations, giving QZ Solutions ground-truth data in a form ready for algorithm validation.
With this dataset, QZ Solutions could test SoilEO’s predictions against laboratory measurements and refine its models, which markedly improved their ability to detect soil nutrients. The partnership continues: QZ Solutions and agrifoodTEF are now testing SoilEO’s fertilisation maps under real-world field conditions.
