Benchmark Tests of Atom Segmentation Deep Learning Models with a Consistent Dataset

Jingrui Wei; Ben Blaiszik; Aristana Scourtas; Dane Morgan; Paul M Voyles

doi:10.1093/micmic/ozac043

Benchmark Tests of Atom Segmentation Deep Learning Models with a Consistent Dataset

Microsc Microanal. 2023 Apr 5;29(2):552-562. doi: 10.1093/micmic/ozac043.

Authors

Jingrui Wei¹, Ben Blaiszik^{2

3}, Aristana Scourtas^{2

3}, Dane Morgan¹, Paul M Voyles¹

Affiliations

¹ Department of Materials Science and Engineering, University of Wisconsin-Madison, Madison, WI, USA.
² Globus, University of Chicago, Chicago, IL, USA.
³ Data Science and Learning Division, Argonne National Laboratory, Lemont, IL, USA.

PMID: 37749717
DOI: 10.1093/micmic/ozac043

Abstract

The information content of atomic-resolution scanning transmission electron microscopy (STEM) images can often be reduced to a handful of parameters describing each atomic column, chief among which is the column position. Neural networks (NNs) are high performance, computationally efficient methods to automatically locate atomic columns in images, which has led to a profusion of NN models and associated training datasets. We have developed a benchmark dataset of simulated and experimental STEM images and used it to evaluate the performance of two sets of recent NN models for atom location in STEM images. Both models exhibit high performance for images of varying quality from several different crystal lattices. However, there are important differences in performance as a function of image quality, and both models perform poorly for images outside the training data, such as interfaces with large difference in background intensity. Both the benchmark dataset and the models are available using the Foundry service for dissemination, discovery, and reuse of machine learning models.

Keywords: atomic scale; image segmentation; neural networks; scanning transmission electron microscopy (STEM).