Researchers from the Massachusetts Institute of Technology have developed a technique to determine material properties, like stress and strain, based on an image of the material showing its internal structure.
It used a machine learning technique called a Generative Adversarial Neural Network that was trained with thousands of paired images — one depicting a material's internal microstructure subject to mechanical forces and the other depicting that same material's color-coded stress and strain values.
What do you think? Do you think internal imaging of materials could become a practical design tool?
Share your questions and comments on our Question of the Week page.