AI support in metal analysis
Our AI models for materials science rely on intelligent image recognition technologies that enable time-efficient and highly precise evaluations of complex metal structures. We rely on complete solutions that are individually tailored to customer requirements.
Yann LeCun, the 2018 Turing Award winner, aims to create AI in computer vision that emulates human understanding. We share this goal, implementing it in our materials science AI models. Instead of annotating countless images, we use intelligent models trained without supervision, yielding smarter results in less time.
Our tests on aluminum polish images, focusing on MgSi and AlFeSi, show promising results. Our partner, HAI (Hammerer Aluminium Industries), supports extensive testing of our solution.
Going beyond segmentation, we automatically classify polishes using various parameters.
Our aim is to develop efficient AI models that reduce annotation time. A versatile optical AI can be applied across different areas, from quality assurance to traffic monitoring. Our focus is on practical, impactful advancements in AI.
In this video, the activations of some layers of a neural network are shown graphically. The activations can be visualized as the "firing" of the neurons. This method provides deeper insights into the behavior of the network.
