Large dataset and benchmarks advance electron microscopy image analysis
A benchmark dataset and baseline methods for four-dimensional STEM diffraction patterns
Computer Vision and Pattern Recognition
Summary
Making sense of the many images produced by advanced electron microscopes is hard because each image is actually a complex pattern showing atomic arrangements. The authors gathered over 170,000 such patterns from experiments and simulations, providing labels and metadata to help computers learn how to interpret them. They also tested methods to detect important features in these patterns and to compare simulated and real data. This collection aims to help build better tools for studying materials at the atomic scale.
What this means in practice
- •For materials scientists: Use labeled 4D-STEM diffraction patterns to improve automated analysis of material structures at the atomic scale.
- •For machine learning engineers: Train and evaluate models for diffraction pattern feature detection and pattern retrieval with a high-volume, quality-ranked dataset combining experiments and simulations.
Authors
Yuyan Guan, Haoran Zhang, Zian Mao, Antong Yang, Caifei Li, Jialong Wang, Chuying Ouyang, Hong Wang, Xiaoqin Zeng, Yujun Xie
Abstract
Four-dimensional scanning transmission electron microscopy (4D-STEM) records a two-dimensional diffraction pattern at each electron-probe position, yielding spatially resolved reciprocal-space information but large, heterogeneous data volumes. Here we describe 4D-ImageNet, a collection of 174,000 diffraction patterns comprising 145,000 experimental patterns selected from 29 acquisitions and 29,000 multislice simulations. The experimental data cover acquisition-level labels for Ag, Au, mixed Au-Ag, CoO, Pd and ZnO specimens across multiple fields of view, scan dimensions, camera lengths and exposure times. Each acquisition contributes 5,000 quality-ranked patterns with source scan coordinates and acquisition metadata. A set-prediction detector provides model-derived pseudo-labels for the direct-beam position and Bragg-disk centres, with a confidence score for each disk. The simulation data cover 13 crystal structures and include Euler rotations, reciprocal-space sampling and approximate low-index beam directions. A grouped mixed-domain masked-reconstruction benchmark is provided to assess leakage-resistant loading and evaluation across experimental and simulated data. The dataset is intended for representation learning, disk detection, diffraction-pattern retrieval, orientation analysis and simulation-to-experiment studies.