JANO

The welding of metallic materials is a task of enormous importance in the world of aerospace engineering, especially in those parts that require high stresses and efforts where the use of X-ray inspections is usually resorted to detect possible defects. This inspection requires an expert who, through different levels of control, carries out the validation of the same, investing in this phase numerous hours of work and staff training.

Therefore, in the present work, we seek to define an algorithm for the detection and classification of defects in industrial X-ray images based on a combination of Computer Vision techniques and Artificial Intelligence techniques based on Machine Learning and Deep Learning. The objective of this algorithm is to improve the inspection process, providing the human inspector with a support system to assist the final detection and classification decision, but with the ability to carry out automatic inspections autonomously if the said system reaches the same Probability of Detection as a human inspector.