{"title":"GeGra:利用深度学习从材料显微镜数据中获取定量粒度分析的通用模型","authors":"","doi":"10.1016/j.matchar.2024.114379","DOIUrl":null,"url":null,"abstract":"<div><p>Grain size has a significant impact on the properties of materials, and is crucial for predicting material properties. Traditional grain size measurement relies heavily on human operators, leading to subjective results, and existing machine learning methods are typically material-specific, requiring significant labeling and training efforts for each new material. This paper provides insight into developing a deep learning-based generic grain boundary detection model (GeGra) from different material micrographs. The model is trained on 1006 images from various microscopy techniques such as light optical, Kerr, and scanning electron microscopy, acquired at different magnifications for different materials such as copper, austenite, brass, sintered hard magnet, hard metal, bronze, nickel silver, and aluminum. The developed GeGra model effectively handles visual artifacts and substructures such as twin grains, which often pose challenges for material-specific, state-of-the-art grain boundary segmentation models. The developed model achieved an IoU score of 69 % on a diverse test set and enables accurate grain size analysis using external image analysis software in less than one minute, according to ASTM standards, which is more than 5 times faster than the manual method. The developed model prioritizes generality with objective that it can have broader applicability for various materials instead of high-precision grain boundary detection. Additionally, the model has the potential to be a foundational tool for generalized grain size analysis in material microscopy, reducing the effort required for such analysis and assisting both material science experts and machine learning users.</p></div>","PeriodicalId":18727,"journal":{"name":"Materials Characterization","volume":null,"pages":null},"PeriodicalIF":4.8000,"publicationDate":"2024-09-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.sciencedirect.com/science/article/pii/S1044580324007605/pdfft?md5=1760ceaf0ca3ed364c3e296bcdb18057&pid=1-s2.0-S1044580324007605-main.pdf","citationCount":"0","resultStr":"{\"title\":\"GeGra: Approaching a generic model for quantitative grain size analysis from materials microscopy data using deep learning\",\"authors\":\"\",\"doi\":\"10.1016/j.matchar.2024.114379\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<div><p>Grain size has a significant impact on the properties of materials, and is crucial for predicting material properties. Traditional grain size measurement relies heavily on human operators, leading to subjective results, and existing machine learning methods are typically material-specific, requiring significant labeling and training efforts for each new material. This paper provides insight into developing a deep learning-based generic grain boundary detection model (GeGra) from different material micrographs. The model is trained on 1006 images from various microscopy techniques such as light optical, Kerr, and scanning electron microscopy, acquired at different magnifications for different materials such as copper, austenite, brass, sintered hard magnet, hard metal, bronze, nickel silver, and aluminum. The developed GeGra model effectively handles visual artifacts and substructures such as twin grains, which often pose challenges for material-specific, state-of-the-art grain boundary segmentation models. The developed model achieved an IoU score of 69 % on a diverse test set and enables accurate grain size analysis using external image analysis software in less than one minute, according to ASTM standards, which is more than 5 times faster than the manual method. The developed model prioritizes generality with objective that it can have broader applicability for various materials instead of high-precision grain boundary detection. Additionally, the model has the potential to be a foundational tool for generalized grain size analysis in material microscopy, reducing the effort required for such analysis and assisting both material science experts and machine learning users.</p></div>\",\"PeriodicalId\":18727,\"journal\":{\"name\":\"Materials Characterization\",\"volume\":null,\"pages\":null},\"PeriodicalIF\":4.8000,\"publicationDate\":\"2024-09-14\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"https://www.sciencedirect.com/science/article/pii/S1044580324007605/pdfft?md5=1760ceaf0ca3ed364c3e296bcdb18057&pid=1-s2.0-S1044580324007605-main.pdf\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Materials Characterization\",\"FirstCategoryId\":\"88\",\"ListUrlMain\":\"https://www.sciencedirect.com/science/article/pii/S1044580324007605\",\"RegionNum\":2,\"RegionCategory\":\"材料科学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q1\",\"JCRName\":\"MATERIALS SCIENCE, CHARACTERIZATION & TESTING\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Materials Characterization","FirstCategoryId":"88","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S1044580324007605","RegionNum":2,"RegionCategory":"材料科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"MATERIALS SCIENCE, CHARACTERIZATION & TESTING","Score":null,"Total":0}
GeGra: Approaching a generic model for quantitative grain size analysis from materials microscopy data using deep learning
Grain size has a significant impact on the properties of materials, and is crucial for predicting material properties. Traditional grain size measurement relies heavily on human operators, leading to subjective results, and existing machine learning methods are typically material-specific, requiring significant labeling and training efforts for each new material. This paper provides insight into developing a deep learning-based generic grain boundary detection model (GeGra) from different material micrographs. The model is trained on 1006 images from various microscopy techniques such as light optical, Kerr, and scanning electron microscopy, acquired at different magnifications for different materials such as copper, austenite, brass, sintered hard magnet, hard metal, bronze, nickel silver, and aluminum. The developed GeGra model effectively handles visual artifacts and substructures such as twin grains, which often pose challenges for material-specific, state-of-the-art grain boundary segmentation models. The developed model achieved an IoU score of 69 % on a diverse test set and enables accurate grain size analysis using external image analysis software in less than one minute, according to ASTM standards, which is more than 5 times faster than the manual method. The developed model prioritizes generality with objective that it can have broader applicability for various materials instead of high-precision grain boundary detection. Additionally, the model has the potential to be a foundational tool for generalized grain size analysis in material microscopy, reducing the effort required for such analysis and assisting both material science experts and machine learning users.
期刊介绍:
Materials Characterization features original articles and state-of-the-art reviews on theoretical and practical aspects of the structure and behaviour of materials.
The Journal focuses on all characterization techniques, including all forms of microscopy (light, electron, acoustic, etc.,) and analysis (especially microanalysis and surface analytical techniques). Developments in both this wide range of techniques and their application to the quantification of the microstructure of materials are essential facets of the Journal.
The Journal provides the Materials Scientist/Engineer with up-to-date information on many types of materials with an underlying theme of explaining the behavior of materials using novel approaches. Materials covered by the journal include:
Metals & Alloys
Ceramics
Nanomaterials
Biomedical materials
Optical materials
Composites
Natural Materials.