Non-destructive testing (NDT) has always been part of TEC Eurolab\u2019s DNA: visual inspections, ultrasonics, radiography, industrial tomography, liquid penetrant testing, magnetic particle testing\u2026 <\/p>\n\n
Over time, we have come to understand the real challenges of industrial quality control: we know what it means to interpret a radiographic image of a casting, as well as a CT scan of an aerospace component; we understand the responsibility that comes with making a judgment that must be accurate, documentable, and traceable; we have experienced the effect of fatigue on an operator who analyzes hundreds of images in sequence; we understand the pressures of production cycle times.<\/p>\n\n
Based on this understanding and our hands-on experience, a question arose: Is it possible to simplify and automate these processes?<\/strong> The first answer came in 2019, when we began exploring the potential of applying artificial intelligence solutions to industry\u2014at a time when AI was known only to industry insiders. This investigation yielded results, starting with one of our most advanced NDT departments: industrial tomography. <\/p>\n\n When we began applying computer vision algorithms to industrial tomography images in 2019, AI was not yet as advanced or widespread as it is today. We did this out of curiosity about the technologies that were emerging in the field of computer vision\u2014technologies that seemed to streamline our work\u2014and, as an NDT laboratory, because we needed to develop our own perspective on artificial intelligence applied to industrial inspection. <\/p>\n\n Today, Sideius, through its BlueTensor division, designs and implements systems based on computer vision and machine learning for automated quality control, defect classification, and operator support. These systems are built on a deep understanding of the context: the types of defects, relevant regulations, acceptable tolerances, the specific challenges of each inspection method, and the problems that client companies face on a daily basis<\/strong>. <\/p>\n\n According to Grand View Research, the global NDT market was valued at $21.28 billion in 2024 and is projected to reach $45.97 billion by 2033, with a CAGR of 9.2% [1<\/a>]. A significant portion of this growth is driven by the integration of artificial intelligence. The specific segment of robotic and drone-based NDT is already worth $1.24 billion in 2025 and will grow to $2.52 billion by 2030, with a CAGR of 15.24% [2<\/a>]. <\/p>\n\n The most telling sign, however, comes from the international scientific community. The International Conference on Non-Destructive Testing for Next Generation 2027 [<\/strong>3<\/strong><\/a>]<\/strong>, to be held in Kobe, Japan, has identified the smart transformation of NDT as its central theme: the incorporation of machine learning, generative AI, digital tools, robots, and drones into inspection processes, along with the push toward automation, labor savings, and the development of new standards and regulations. <\/p>\n\n Deep learning models applied to inspection make it possible to automate repetitive, monotonous, and requiring a high degree of precision\u2014such as in the case of surface defects on steel, where accuracy levels have exceeded 96%, with the best model reaching 99.26%, outperforming both traditional rule-based software and human inspection [4<\/a>].<\/p>\n\n When it comes to robots and drones used for in-the-field NDT, particularly in situations that are hazardous to humans, the range of applications is expanding:<\/p>\n\n At Sideius, we combine our specialized expertise in the field of NDT with the cross-disciplinary nature of machine learning and deep learning to support or automate quality control processes both internally and at our clients\u2019 sites.<\/strong><\/p>\n\n Some applications:<\/p>\n\n 1. We use AI to enhance the analysis of tomographic images for the classification of porosity, inclusions, and cracks in additive manufacturing components or die-cast parts:<\/p>\n\nThe Growth of the NDT Industry Worldwide, driven by AI<\/h2>\n\n
\n
Concrete Examples of Application in Sideius<\/h2>\n\n