Finding the Unknown: Real-Time Foreign-Object Detection with Hyperspectral AI

Many industrial inspection tasks fail not because images cannot be acquired, but because reliable decisions must be made under real production constraints. This is particularly true for foreign-object detection and anomaly inspection, where defects are rare, highly variable, or not known in advance.

This presentation explores how hyperspectral imaging combined with modern AI methods can enable robust real-time inspection workflows beyond the limitations of conventional RGB-based machine vision. Using examples from industrial food inspection and sorting, the talk examines how spectral information changes the representation of inspection problems and improves the detection of visually ambiguous or previously unseen objects.

The session focuses on practical workflow design - from snapshot hyperspectral acquisition and spectral preprocessing to anomaly detection, decision pipelines, and deployment considerations in production environments. Particular attention is given to approaches that reduce dependency on large labeled defect datasets and allow systems to adapt to changing materials and inspection conditions.

Rather than presenting hyperspectral imaging as simply “more data,” the talk discusses how spectral and AI-based methods can fundamentally change the way industrial inspection systems are designed, trained, and scaled for real-world operation. The session is for system integrators, machine vision engineers, and industrial AI developers interested in moving beyond conventional RGB inspection toward robust real-time anomaly detection and decision-making workflows.

About the presenter
Marie-Theres Heine, Ph.D., is the head of sales at Cubert GmbH, a leader in real-time spectral imaging. After a decade in industrial automation, Heine joined the company in 2024, putting the focus on machine vision. She holds a doctorate in physics from the University of Ulm, where she graduated magna cum laude, after completing her diploma thesis in econophysics. Heine’s deep expertise in physics and her vision for the use of hyperspectral imaging in industrial applications drive Cubert’s growth and success in this emerging market.