Speaker
Description
Recent advances in Artificial Intelligence (AI) have helped researchers of various fields develop fast, intelligent and reliable tools which can automate manual processes, save time and reduce user bias in the results. We present two real-world use cases of Artificial Intelligence and Computer Vision which were proven to provide trustfull results. The first use case refers to a methodology for the automated analysis of metal powder SEM images towards material characterization. This tool takes advantage of a combination of recent Artificial Intelligence advances (Faster R-CNN), conventional Image Processing techniques and characterization prior knowledge to assess metal powder quality for Additive Manufacturing applications. SEM is being used for characterizing metal powder, specifically diameter and amount of spherical structures, which are key characteristics for assessing the quality of the analyzed powder. We propose a method that can overcome the above challenges and is also developed as a software. Initially, a Deep Neural Network detects spherical structures with structural malformations which reduce quality. Subsequently, conventional circle detection techniques are applied to detect perfectly spherical structures. Finally, the remainder structures are further analyzed with custom classifiers. All the above information is stored and reported to the user as a final statistical result giving an extensive report on the distribution of structural types that were identified in the material. In addition, the proposed method can be extended to accurate quality assessment. This can be done with a semi-supervised way using the aforementioned collected data combined with one-time user input. The second use case is a software which uses a custom made algorithm which detects Carbon Nanotubes in SEM images and estimates various statistics on the diameter distribution. Both use cases highlight the advantages of using newly developed algorithms of Artificial Intelligence and Image Processing towards advanced and automated characterization.
| Speaker Country | Greece |
|---|