QualiCrop consortium brings earlier defect detection and robotic sorting at commercial packhouse speeds

QualiCrop, an Innovate UK-funded project led by Fotenix with Leap AI and the University of Lincoln, combines multispectral quality inspection with robotic handling to identify defects, including those not yet visible to the human eye, and remove affected tomatoes at commercial packhouse speeds.

A consortium of agri-tech companies and researchers built an automated grading line that finds defective tomatoes and picks them off a moving belt. The system can automatically separate healthy fruit from defective fruit with 87% accuracy. 

Most automated packhouse sorters rely primarily on RGB cameras, which assess characteristics already visible on the fruit’s surface. The QualiCrop line uses Fotenix’s Automated Quality Control (AQC), a multispectral system that analyses reflected light across targeted wavelengths to identify changes associated with damage, disease and mould before visible symptoms develop. In commercial settings, it routinely grades and classifies up to 4,000 items a minute. 

Existing optical sorting systems can cost upwards of £1million and often require a bespoke installation. QualiCrop aimed to make advanced automated grading more affordable for small and medium-sized UK greenhouse businesses operating across three to 25 hectares. It paired Fotenix AQC with Leap AI’s intelligent robotic system, which can precisely locate and gently handle fruit at production speeds. 

Charles Veys, CEO of Fotenix, explained: “The AQC classified each tomato in real-time so the robotic sorting cell could act on that data while the fruit was still moving. That quality data was also collated by batch and growing site, delivering insight growers can use to help to inform growing strategies upstream.”

Ben Stuart, CEO of Leap AI, commented: “As specialists in the robotic handling of fresh produce, we are always looking to bring additional value to our customers. We are excited to pair our top-of-the-line automation, including defect removal with our existing end-of-line packing solutions.”

The project was funded through the Farming Innovation Programme, part of Defra’s investment in innovation, research and development, delivered in partnership with Innovate UK, part of UK Research and Innovation. The University of Lincoln developed detection algorithms that reduced the data required to identify a defect from 400 wavelengths to 18, helping inspection operate at a speed and cost appropriate for a commercial packhouse.

Fotenix led the project and supplied the multispectral imaging and real-time classification technology. The University of Lincoln hosted the commercial trials and UK grower APS Salads supplied the tomatoes used to develop and test the integrated system. 

QualiCrop has helped inform further commercial activity. Fotenix AQC is commercially deployed and grading and classifying produce within 0.3 seconds, while Leap AI has built specialist hardware and software for end-of-line packing solutions for fresh produce.

On the research side, the project has advanced the University of Lincoln’s expertise in spectral imaging, translating research into practical industry applications. It delivered new imaging hardware, a unique spectral tomato dataset, anomaly characterisation methods and efficient machine-learning algorithms. The project also strengthened industry collaboration and upskilled junior researchers, supporting their progression into R&D careers. The University is taking the imaging research into two further Innovate UK projects focused on tomato harvesting and glasshouse growing systems.

• : *I have permission from the copyright holder to publish this content and images.