2020

Xpectrum

Development
Design
Architecture

task

Ensuring food safety, guaranteeing high-quality and fair food products, that meet all regulations, is nowadays a very expensive and time and labor-intensive process. The goal of Xpectrum is to ensure quality, safety and authenticity in food retail in a reliable, quick and easy to operate manner by applying NIR spectroscopy. This way Xpectrum enables retailers to test substantially more samples than currently in a routine setting. As a result, more consumer trust will be gained by providing more transparency on the quality of their food. Xpectrum is a product developed with support from the Pure and Sure subsidy project by the province of Vlaams-Brabant.

idea

Xpectrum developed a scalable approach that enables retailers to screen food samples for quality issues and fraud on a large scale in a cost-efficient way. It therefore uses NIR spectroscopy coupled with cloud technologies and advanced machine learning. Food retailers will be able to obtain results anywhere, from the reception of the goods at the distribution center, in the field, or during site visits... Results will be shown immediately so that the specific escalation processes of each product can be launched immediately.

approach

In NIR spectroscopy, a substance is put on a NIR spectrometer and exposed to a broad-spectrum of near infrared light, which can be absorbed, transmitted, reflected or scattered by the sample of interest. The light intensity as a function of wavelength is measured before and after interacting with the sample, and the diffuse reflectance, a combination of absorbance and scattering, caused by the sample is calculated. The resulting patterns are sample specific. Using artificial intelligence, a model is designed that is able to predict the quality or authenticity of the food sample. Anomalies in the product composition, such as water, protein, fat etc can be detected.

xpectrum infographic of how xpectrum works

technology

XAOP is glad to announce that we were a part of the full application development. For this project XAOP developed the cloud infrastructure that manages all the data and prediction models. We provided technical architecture and solid back-end development and created the processes for data gathering and processing. In addition, a framework was put into place that enables efficient use of machine learning capabilities. Furthermore, an easy to use front-end application oriented towards operators in industry has been developed by the team which facilitates metadata gathering and connectivity with the (Bluetooth) NIR spectrometer.