| Abstract Scope |
This work reviews the latest advancements in AI-driven microstructure characterization of titanium alloys. By uniquely integrating classical computer vision, custom-trained deep learning convolutional neural networks, and locally deployed large vision zero-shot transformer models, highly challenging microstructure characterization tasks are now able to be fully automated. This includes, but is not limited to, individual alpha lath discretization and quantification in near-alpha alloys, colony versus basketweave morphology classification, colony segmentation, and equiaxed-alpha cluster separation. These capabilities enhance the fundamental understanding of structure-property-processing relationships, crucial for alloy design and processing strategies, and this work aligns with the symposium's focus on microstructure exploration and performance evaluation in titanium and its alloys. |