The main result of the AIQAS project was the development of a system that allows multi-defect and multi-material inspection of pieces produced by continuous production line type industrial processes.


Detect, in a single station, different defects and anomalies in real time on different materials.

Getting the system to incremental and flexible learning through semi-supervised labelling of anomalies detected.

Deploy IA models on computing hardware physically located on the production line (IoT Edge Computing).


The developed solution is based on real-time analysis of multiple defect types simultaneously, with consequent cost reduction, and implementable on different flat-type materials. Using deep learning techniques (convolutional neural networks), the image can be treated holistically and automatically learn the most effective features to characterise the defects, without the need for pre-specification. This is a major advance over existing solutions, which consist of a separate, material-specific analysis of different types of defects.

Participating entities

Funding Entity

  • Institute for the Economic Development of the Principality of Asturias


  • Izertis


  • Idonial Foundation