About this Abstract |
Meeting |
2021 TMS Annual Meeting & Exhibition
|
Symposium
|
Practical Tools for Integration and Analysis in Materials Engineering
|
Presentation Title |
Tools for Microstructural Analysis Using Computer Vision and Machine Learning |
Author(s) |
Elizabeth A. Holm, Bo Lei, Andrew Kitahara, Nan Gao, Ryan Cohn |
On-Site Speaker (Planned) |
Elizabeth A. Holm |
Abstract Scope |
Microstructural science relies on quantitative characterization and analysis of microstructure, typically based on images obtained from microscopy, diffraction, or other experimental modalities. Recent progress in data science, including computer vision (CV) and machine learning (ML), offer new approaches to extracting information from microstructural images. However, because they are generally developed to analyze natural images, and because they are not part of the materials curriculum, applying these methods to materials problems is not always straightforward. This talk presents the basic steps for encoding and analyzing microstructural image data: image preprocessing and data augmentation; feature vector construction; and unsupervised and supervised machine learning. A Python code tutorial that applies these operations on an open access data set of steel defects is included. Case studies demonstrate practical aspects of developing CV/ML workflows, including dataset considerations, hyperparameter selection, ML technique, and potential pitfalls. |
Proceedings Inclusion? |
Planned: |
Keywords |
Machine Learning, Characterization, Computational Materials Science & Engineering |