Abstract
Cutting edge deep learning techniques allow for image segmentation with great speed and accuracy. However, application to problems in materials science is often difficult since these complex models may have difficultly learning meaningful image features that would enable extension to new datasets. In situ electron microscopy provides a clear platform for utilizing automated image analysis. In this work, we consider the case of studying coarsening dynamics in supported nanoparticles, which is important for understanding, for example, the degradation of industrial catalysts. By systematically studying dataset preparation, neural network architecture, and accuracy evaluation, we describe important considerations in applying deep learning to physical applications, where generalizable and convincing models are required. With a focus on unique challenges that arise in high-resolution images, we propose methods for optimizing performance of image segmentation using convolutional neural networks, critically examining the application of complex deep learning models in favor of motivating intentional process design.
Highlights
In situ and operando experimental techniques, where dynamic process can be observed with high temporal and spatial resolution, have allowed scientists to observe chemical reactions, interfacial phenomena, and mass transport processes to give a better understanding of the physics of materials phenomena, and a view into how materials react under the conditions in which they are designed to perform[1,2]
Building on previous work on image segmentation, automated analysis, and merging deep learning within the field of materials science, we study a variety of convolutional neural networks (CNNs) architectures to define the most important aspects for the practical application of deep learning to our task
Our results suggest that shallow, wide CNNs have enough expressive power to segment high-resolution image data[32]
Summary
In situ and operando experimental techniques, where dynamic process can be observed with high temporal and spatial resolution, have allowed scientists to observe chemical reactions, interfacial phenomena, and mass transport processes to give a better understanding of the physics of materials phenomena, and a view into how materials react under the conditions in which they are designed to perform[1,2]. The high dimensionality of data at inter- nating encoded and decoded images of the same resolution mediate layers of a neural network combined with the compound followed by a single convolutional layer and activation function to connections between hidden layers makes representation, and relate unique aspects of both images These challenges— representation and visualization of CNN models—are areas of active research[17,18]
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
