LaueDB: A Dataset for Laue Patterns

Štěpán Venclík, Tomáš Červeň, Jan Kříž, David Sviták, Petr Čermák

Charles University, Prague, Czech Republic

 

Laue diffraction is a widely used technique for orienting single crystals and a routine procedure during sample preparation for many scientists. Over the years, a variety ofsoftware tools have been developed to assist in interpreting Laue patterns [1, 2]. Despite significant progress in image processing and pattern recognition, a robust and fully automated solution for indexing Laue patterns has yet to be achieved.

In recent years, machine learning has emerged as a promising approach to tackle this challenge [3]. However, the development and validation of more advanced algorithms are currently hindered by the lack of annotated experimental datasets. As a result, all training and testing are still conducted exclusively on synthetic data.

LaueDB aims to bridge this gap by creating a dataset of oriented X-ray and neutron Laue patterns that could serve as a training and evaluation dataset for both classical and machine learning approaches.

We plan to utilise the Automatic Laue Sample Aligner (ALSA) [4] to create the initial dataset, capturing a large number of patterns for each sample crystal, as well as collaborate with research infrastructures to develop a submission pipeline for patterns created during routine sample orientation. In addition, existing tools and algorithms for peak finding and Laue indexing will be compared.

1.         Esmeralda Laue Suite (https://code.ill.fr/scientific-software/esmeralda)

2.         Clip4 (https://clip4.sourceforge.net/)

3.         Purushottam Raj Purohit, R. R. P., Tardif, S., Castelnau, O., Eymery, J., Guinebretiere, R., Robach, O., Ors, T. & Micha, J.-S. (2022). J. Appl. Cryst. 55, 737-750.

4.         ALSA (https://charlesautomata.cz/alsa)