Catalogue of Services

Technical evaluation of frameworks developed for the creation of virtual assistants, focusing on elements such as the underlying architecture, natural language processing methods, and interactions with databases. Tests will include measuring response times, the ability to handle simultaneous queries, and overall system efficiency. Potential security vulnerabilities will also be identified, with recommendations proposed to strengthen data protection.

Test design

Lab gold-standard evaluation

This service provides a lab-based evaluation for AI classification models, contrasting their results with a gold-standard analysis to assess accuracy and reliability. Focused on applications in food processing, such as beekeeping and pollen counting, this service ensures that AI models meet high standards of precision, which is essential for fields requiring validated data accuracy.

Lab testing and validation of non-destructive spectral devices focused on agri-food products

This service supports companies that are developing non-destructive spectral (NDS) instrumentation for agri-food applications, providing validation studies focused on spectral repeatability, signal-to-noise ratio, and quality assessment. Using a well-equipped sensor lab, the service includes comparisons with commercially available instruments and access to a referenced agri-food sample bank. By conducting repeatability and performance assessments, clients gain insights into device robustness and potential design improvements.

Conformity assessment

Performance testing of deep learning solutions

This service evaluates the performance of deep learning algorithms to determine their efficiency and feasibility for migration to embedded devices. By leveraging high-performance computing (HPC) and GPU resources, this testing helps developers understand how well their algorithms perform in terms of speed, accuracy, and resource consumption. The service is particularly beneficial for applications in food processing, such as beekeeping and pollen counting, where real-time data processing is essential.

Testing and validation of the performance of predictive models based on nondestructive spectral sensor data (NIR and hyperspectral data).

This service validates the performance of multivariate predictive models that utilise nondestructive spectral sensor data, such as NIR and hyperspectral imaging. Through statistical and chemometric analysis, the service assesses model accuracy and provides recommendations for optimization. Clients benefit from access to a comprehensive software suite dedicated to chemometrics and statistical studies, as well as options for database transfer to adapt models across devices. The service also includes access to a large agri-food sample bank for model evaluation and refinement.

Conformity assessment

Validation of an intelligent microscope for automated honey quality analysis through corbicular pollen analysis

Location

Spain

Food processing

The validation of the service will be carried out using corbicular pollen samples, which will be examined through the established traditional unitary methodology and with an AI-equipped microscope. The results will be statistically compared to determine the error rate of the AI in the analysis.

Validation of an intelligent microscope for automated honey quality analysis through pollen analysis

Location

Spain

Food processing

The validation of the service will be carried out using honey samples, which will be examined through the established traditional unitary methodology and with an AI-equipped microscope. The results will be statistically compared to determine the error rate of the AI in the analysis.