当前位置:首页 >期刊论文 >《最新论文》>正文

首个带注解的纳米科学SEM图像库

 2018/10/12 10:51:07 《最新论文》 作者:Scientific Data 我有话说(0人评论) 字体大小:+

论文标题:The first annotated set of scanning electron microscopy images for nanoscience

期刊:Scientific Data

作者:Rossella Aversa, Mohammad Hadi Modarres, Stefano Cozzini, Regina Ciancio & Alberto Chiusole

发表时间:2018/08/28

数字识别码:10.1038/sdata.2018.172

原文链接:https://www.nature.com/articles/sdata2018172?utm_source=Other_website&utm_medium=Website_links&utm_content=RenLi-MixedBrand-multijournal-Multidisciplinary-China&utm_campaign=ORG_USG_JRCN_RL_article_promotion_sciencenet_Oct_2nd

数据获取在研究中发挥着越来越重要的作用。实际上,许多研究领域几乎完全依赖于开放获取且管理完备的全球共享数据库,这些功能目前大多由数据存储平台实现,因而数据存储平台已成为科学研究基础设施的重要组成部分。

在此背景下,致力于欧洲纳米科学研究的NFFA-EUROPE项目(www.nffa.eu)将设计欧洲范围内有效的纳米级数据共享方法确立为其主要任务之一。该项目在欧洲共有20个合作单位,以及150多种不同的实验/计算仪器和技术。项目活动产生的科学数据将存储于NFFA-EUROPE信息和数据存储平台(IDRP)。该平台是完全开放的,科研人员在遵守数据政策的前提下可自由获取平台上的数据。IDRP还配套有一系列数据分析服务,这些服务自NFFA-EUROPE项目开始以来就在不断进化。

在NFFA-EUROPE项目的设备中,扫描电子显微镜(SEM)是最常用的仪器之一,10个NFFA-EUROPE站点都配有SEM。SEM是一种常规使用的表征技术,它通过将聚焦电子束扫描到样品表面上,以提供样品的形貌和组成信息,其分辨率可达纳米级别。

第一个实现的数据分析是由NFFA-EUROPE IDRP提供的服务,其核心是一个可自动进行图像识别的工具,可用于帮助存储、分类和标记SEM图像:我们采用监督式机器学习算法,使用深度卷积神经网络识别SEM图像。为对网络进行训练,我们必须提供已标记的训练集,即一组已由人正确分类的SEM图像。

在《科学数据》发表的The first annotated set of scanning electron microscopy images for nanoscience一文中,来自CNR-IOM材料研究所的Rossella Aversa及同事建立了第一个公开的人类注解的扫描电子显微镜(SEM)图像数据集。大约26,000张纳米SEM图像被划分为10个类别,进而分别纳入4个适合于图像识别任务的标注训练组。这10个类别包括零维物质如粒子、一维物质如纳米线和纤维、二维物质如薄膜、涂层表面以及有图案表面,三维结构如微机电系统(MEMS)器件和柱结构等。类别中还包括小部件、生物结构等以尽可能扩展图像范围。通过为各个类别创建子树结构,并将可用的图像尽可能归入所属类别,从而为该图像数据集引入了初步的层次结构。

图1:SEM数据集各类别的代表图像。

摘要:In this paper, we present the first publicly available human-annotated dataset of images obtained by the Scanning Electron Microscopy (SEM). A total of roughly 26,000 SEM images at the nanoscale are classified into 10 categories to form 4 labeled training sets, suited for image recognition tasks. The selected categories span the range of 0D objects such as particles, 1D nanowires and fibres, 2D films and coated surfaces as well as patterned surfaces, and 3D structures such as microelectromechanical system (MEMS) devices and pillars. Additional categories such as tips and biological are also included to expand the spectrum of possible images. A preliminary degree of hierarchy is introduced, by creating a subtree structure for the categories and populating them with the available images, wherever possible.

阅读论文全文请访问:https://www.nature.com/articles/sdata2018172?utm_source=Other_website&utm_medium=Website_links&utm_content=RenLi-MixedBrand-multijournal-Multidisciplinary-China&utm_campaign=ORG_USG_JRCN_RL_article_promotion_sciencenet_Oct_2nd

期刊介绍:Scientific Data (https://www.nature.com/sdata/) is a peer-reviewed, open-access journal for descriptions of scientifically valuable datasets, and research that advances the sharing and reuse of scientific data. Scientific Data welcomes submissions from a broad range of research disciplines, including descriptions of big or small datasets, from major consortiums to single research groups. Scientific Data primarily publishes Data Descriptors, a new type of publication that focuses on helping others reuse data, and crediting those who share.

The 2017 journal metrics for Scientific Data are as follows:

•2-year impact factor: 5.305

•5-year impact factor: 5.862

•Immediacy index: 0.843

•Eigenfactor® score: 0.00855

•Article Influence Score: 2.597

•2-year Median: 2

来源:Scientific Data