R-AI-VQC: REACH Agrifood AI-supported Visual Quality Control
REACH equips small and mid-size food processors with an always-on, AI-based inspection system that catches defects the moment they appear and feeds live process data back to the line. The goal: keep shelves safe, cut waste and free skilled operators for higher-value work.
Why it matters
Food safety and spotless packaging are non-negotiable, yet skilled quality-assurance staff are increasingly hard to find. Varga-Szárnyas Kft. – the pilot plant behind REACH – must meet rising customer expectations while compensating for this talent gap. Continuous, data-driven inspection is the only scalable answer: by collecting real-time machine and video data on the packaging line, hazards can be spotted and fixed before products ever leave the factory. As the company expands, AI model-training promises continuous quality gains without adding head-count
What we built
REACH installs a digital “inspector” above the conveyor:
- High-definition cameras stream every pack to the cloud.
- Deep-neural-network models, refined through transfer learning, flag the tiniest flaw within milliseconds.
- Real-time statistical-process-control dashboards show operators exactly where – and when – to intervene, so quality issues are solved at the start of production, not the end.
This combination delivers a cost-effective, SME-friendly alternative to manual spot-checks.
Impact so far
- A 15 % jump in production efficiency at Varga-Szárnyas Kft., thanks to fewer unplanned stops and faster decisions .
- Around 10 % improvement in overall product quality, cutting waste and recall risk .
- A growing cloud-based agrifood image library that lets other SMEs deploy the same AI with minimal extra data collection
What happens next
Mortoff Kft. and Varga-Szárnyas Kft. are now weaving the REACH engine into a full Manufacturing Execution System (MES), while mapping new production hot-spots where AI can lift both throughput and quality . Similar upgrades have already been proposed to other European food processors, and the team will continue expanding the model library so that state-of-the-art visual inspection becomes a plug-and-play service for agrifood SMEs everywhere.
