Vol.26, Special Issue A, 2026, pp. S39–S44
https://doi.org/10.69644/ivk-2026-siA-0039

Integrating artificial intelligence into various manufacturing and quality control processes is extremely popular these days and widely implemented by different companies in the field of automotive, aviation, industry manufacturing as well as pharmaceutical industry. Significantly, the integration of artificial intelligence into material science fields such as metallography and quantitative image analysis techniques has been pivotal in analysing and understanding the microconstituents and microstructures of different metals. Deep learning-based methods for image analysis and segmentation are employed to allocate semantic tags on every pixel of the image and therefore dividing the image into more detailed sections. However, this process is not that simple as it seems at first, since several parameters can be critical in terms of effectiveness. For example, such parameters are the choice of the learning method, as well as the reliability of thousands of already processed image-based information that form the basis of the procedure. Perhaps one of the huge tasks is collecting and organising the right amount of processed image information. This article is about presenting the complexity of the topic, highlighting various learning methods employed in enhancing the AI developed models and the set of information that is essential for providing usable end results, that can be integrated into the process of metallographic and quantitative image analysis.

Keywords: • artificial intelligence • quantitative image analysis • AI based prediction • crack propagation prediction • problem identification