The first annotated set of scanning electron microscopy images for nanoscience

Aversa, Rossella; Modarres, Mohammad Hadi; Cozzini, Stefano; Ciancio, Regina; Chiusole, Alberto

doi:10.1038/sdata.2018.172

Cited by 73 publications

(41 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…While ImageNet encompasses a wide range of classes, the noise characteristics in microscopic images are different. Potentially, pre-training with other data sets exhibiting a smaller domain gap such as miscellaneous nanoscientific objects in SEM 47 or ultra-high carbon steel phases SEM 48 can be advantageous. Variances and generalization.…”

Section: Discussionmentioning

confidence: 99%

A Deep Learning Approach for Complex Microstructure Inference

Durmaz

Müller

Lei

et al. 2021

Preprint

View full text Add to dashboard Cite

Automated, reliable, and objective microstructure inference from micrographs is an essential milestone towards a comprehensive understanding of process-microstructure-property relations and tailored materials development. However, such inference, with the increasing complexity of microstructures, requires advanced segmentation methodologies. While deep learning (DL), in principle, offers new opportunities for this task, an intuition about the required data quality and quantity and an extensive methodological DL guideline for microstructure quantification and classification are still missing. This, along with a lack of open-access data sets and the seemingly intransparent decision-making process of DL models, hampers its breakthrough in this field. We address all aforementioned obstacles by a multidisciplinary DL approach, devoting equal attention to specimen preparation, contrasting, and imaging. To this end, we train distinct U-Net architectures with 30–50 micrographs of different imaging modalities and corresponding EBSD-informed annotations. On the challenging task of lath-bainite segmentation in complex-phase steel, we achieve accuracies of 90% rivaling expert segmentations. Further, we discuss the impact of image context, pre-training with domain-extrinsic data, and data augmentation. Network visualization techniques demonstrate plausible model decisions based on grain boundary morphology and triple points. As a result, we resolve preconceptions about required data amounts and interpretability to pave the way for DL's day-to-day application for microstructure quantification.

show abstract

Section: Discussionmentioning

confidence: 99%

A Deep Learning Approach for Complex Microstructure Inference

Durmaz

Müller

Lei

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…Then, retrain the network allowing back-propagation through all the layers. We adopted checkpoints pre-trained on two data sets: ImageNet [17], a large visual database designed for object recognition, in the version of the ImageNet Large Scale Visual Recognition Challenge (ILSVRC) 2012 [18], and our own SEM data set [11]. We note that the second case cannot be formally defined as a transfer learning technique, since the fine tuning of the CNN is performed on the same data set of the checkpoint; nevertheless, this is a commonly adopted way to efficiently refine the parameters of the network.…”

Section: Supervised Learningmentioning

confidence: 99%

“…Table 1. Number of images for each label in the 1μ-2μ data set, adopting the same labelling used in [11,12,13], reported here for completeness: 0 = Porous sponges, 1 = Patterned surfaces, 2 = Particles, 3 = Films and coated surfaces, 4 = Powders, 5 = Tips, 6 = Nanowires, 7 = Biological, 8 = MEMS devices and electrodes, 9 = Fibres.…”

Section: The 1μ-2μ Data Set Selectionmentioning

confidence: 99%

“…As a fundamental step, a sample counting more than 18,000 images was extracted from the original sample and manually labelled in 10 categories, forming the SEM data set [11], which we employed in [12] and in [13] to investigate transfer learning [14], in particular feature extraction, for automatic image categorization. The test accuracy resulting by the adopted technique settled around 90%.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Deep Learning, Feature Learning, and Clustering Analysis for SEM Image Classification

et al. 2020

Self Cite

View full text Add to dashboard Cite

In this paper, we report upon our recent work aimed at improving and adapting machine learning algorithms to automatically classify nanoscience images acquired by the Scanning Electron Microscope (SEM). This is done by coupling supervised and unsupervised learning approaches. We first investigate supervised learning on a ten-category data set of images and compare the performance of the different models in terms of training accuracy. Then, we reduce the dimensionality of the features through autoencoders to perform unsupervised learning on a subset of images in a selected range of scales (from 1 μm to 2 μm). Finally, we compare different clustering methods to uncover intrinsic structures in the images.

show abstract

“…A typical dataset involves 10 7 –10 10 detected events with a total size of up to a few hundred gigabytes (GBs), depending on the number of coordinates measured (3D or 4D) and the required SNR. Unlike the large 2D or 3D image-based datasets, such as those obtained in various forms of optical 26 , 27 and electron microscopy techniques 28 , 29 , processing and conversion of tabulated single-event data requires additional steps of statistical computing for conversion into standard images. This motivates the current workflow development for efficient data processing and analysis.…”

Section: Introductionmentioning

confidence: 99%

An open-source, end-to-end workflow for multidimensional photoemission spectroscopy

et al. 2020

View full text Add to dashboard Cite

Characterization of the electronic band structure of solid state materials is routinely performed using photoemission spectroscopy. Recent advancements in short-wavelength light sources and electron detectors give rise to multidimensional photoemission spectroscopy, allowing parallel measurements of the electron spectral function simultaneously in energy, two momentum components and additional physical parameters with single-event detection capability. Efficient processing of the photoelectron event streams at a rate of up to tens of megabytes per second will enable rapid band mapping for materials characterization. We describe an open-source workflow that allows user interaction with billion-count single-electron events in photoemission band mapping experiments, compatible with beamlines at 3rd and 4rd generation light sources and table-top laser-based setups. The workflow offers an end-to-end recipe from distributed operations on single-event data to structured formats for downstream scientific tasks and storage to materials science database integration. Both the workflow and processed data can be archived for reuse, providing the infrastructure for documenting the provenance and lineage of photoemission data for future high-throughput experiments.

show abstract

The first annotated set of scanning electron microscopy images for nanoscience

Cited by 73 publications

References 8 publications

A Deep Learning Approach for Complex Microstructure Inference

A Deep Learning Approach for Complex Microstructure Inference

Deep Learning, Feature Learning, and Clustering Analysis for SEM Image Classification

An open-source, end-to-end workflow for multidimensional photoemission spectroscopy

Contact Info

Product

Resources

About