Bringing AI to X-ray food inspection, transforming images and machine data into tools for quality, control, and new digital evolutions.
Rayonics is a company that has been designing, manufacturing, and installing X-ray inspection systems for food quality control since 1986, with over 200 machines installed worldwide. The challenge addressed in the project concerns the evolution of X-ray inspection systems for the food industry: moving from traditional vision logic to solutions capable of interpreting images, anomalies, and process data in a more advanced way. In the food context, machines for product inspection via X-rays play a critical role in ensuring quality and food safety. Within the scope of Rayonics machinery, CIM worked on an evolutionary trajectory aimed at enabling real-time monitoring, predictive maintenance, and contaminant analysis through artificial intelligence algorithms.
The project therefore stems from three closely related industrial needs: improving automatic anomaly recognition in X-ray images, making processing compatible with real-time constraints, and building an architectural roadmap capable of leveraging machine-generated data in an edge-cloud perspective.
CIM supported Rayonics through a structured approach across three areas: business analysis and technology development roadmap, development of a Proof of Concept for anomaly detection on X-ray images, and definition and design of an Edge-Cloud architecture for TO-BE machinery management.
Object detection models based on YOLOv11 and Faster R-CNN were developed and tested for identifying contaminants, deformations, and content anomalies on glass and tin containers. The YOLO models were subsequently optimized through TensorRT for execution on NVIDIA Jetson AGX Orin, making inference compatible with real-time automatic inspection scenarios.
In parallel, we analyzed the existing Rayonics architecture and proposed an edge-cloud evolutionary trajectory, with integration of sensors, edge gateways, data lakes, machine monitoring, and cloud services for remote monitoring and predictive maintenance.
The project enabled validation of a technological foundation for the evolution of Rayonics systems toward more intelligent, autonomous, and scalable machines. The developed models achieved high performance in identifying anomalies on X-ray images following optimization on NVIDIA Jetson AGX Orin hardware. Beyond the Proof of Concept, we supported the company by transferring tools and expertise to generate new synthetic datasets, train new models, test new data, and replicate optimization activities. The Edge-Cloud roadmap defined in the project also offers a concrete direction for enabling real-time monitoring, predictive maintenance, structured data management, and new digital after-sales services.
Bringing AI to X-ray food inspection, transforming images and machine data into tools for quality, control, and new digital evolutions.
Complete the form to receive the Free Preview directly in your inbox.
"*" indicates required fields
"*" indicates required fields
Complete the form and proceed to payment to receive the White Paper directly in your inbox.
"*" indicates required fields