Data augmentation in microscopic images for material data mining

作     者：Boyuan Ma Xiaoyan Wei Chuni Liu Xiaojuan Ban Haiyou Huang Hao Wang Weihua Xue Stephen Wu Mingfei Gao Qing Shen Michele Mukeshimana Adnan Omer Abuassba Haokai Shen Yanjing Su 

作者机构：Beijing Advanced Innovation Center for Materials Genome EngineeringUniversity of Science and Technology BeijingBeijing 100083China School of Computer and Communication EngineeringUniversity of Science and Technology BeijingBeijing 100083China Beijing Key Laboratory of Knowledge Engineering for Materials ScienceBeijing 100083China Institute for Advanced Materials and TechnologyUniversity of Science and Technology BeijingBeijing 100083China School of Materials Science and EngineeringUniversity of Science and Technology BeijingBeijing 100083China School of Materials Science and TechnologyLiaoning Technical UniversityLiaoning 114051China The Institute of Statistical MathematicsResearch Organization of Information and SystemsTachikawaTokyo 190-8562Japan National Intellectual Property AdministrationBeijing 100088China Faculty of Engineering SciencesUniversity of BurundiBujumburaBurundi Faculty of Engineering and TechnologyPalestine Technical University–KadoorieTulkaremPalestine College of Information Science and EngineeringChina University of PetroleumBeijingChina Key Lab of Petroleum Data MiningChina University of PetroleumBeijingChina 

出 版 物：《npj Computational Materials》 (计算材料学（英文）)

年 卷 期：2020年第6卷第1期

页      面：601-609页

核心收录：

学科分类：08[工学] 080203[工学-机械设计及理论] 0805[工学-材料科学与工程（可授工学、理学学位）] 0802[工学-机械工程] 0701[理学-数学] 0801[工学-力学（可授工学、理学学位）] 0702[理学-物理学] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：The authors acknowledge financial support from the National Key Research and Development Program of China(No.2016YFB0700500) the National Science Foundation of China(No.51574027,No.61572075,No.6170203,No.61873299) the Finance science and technology project of Hainan province(No.ZDYF2019009) the Fundamental Research Funds for the University of Science and Technology Beijing(No.FRF-BD-19-012A,No.FRF-TP-19-043A2). 

主　　题：mining instance competitive 

摘      要：Recent progress in material data mining has been driven by high-capacity models trained on large datasets.However,collecting experimental data(real data)has been extremely costly owing to the amount of human effort and expertise required.Here,we develop a novel transfer learning strategy to address problems of small or insufficient data.This strategy realizes the fusion of real and simulated data and the augmentation of training data in a data mining procedure.For a specific task of grain instance image segmentation,this strategy aims to generate synthetic data by fusing the images obtained from simulating the physical mechanism of grain formation and the“image styleinformation in real images.The results show that the model trained with the acquired synthetic data and only 35%of the real data can already achieve competitive segmentation performance of a model trained on all of the real data.Because the time required to perform grain simulation and to generate synthetic data are almost negligible as compared to the effort for obtaining real data,our proposed strategy is able to exploit the strong prediction power of deep learning without significantly increasing the experimental burden of training data preparation.