Project Summary: Computer vision models have become a mainstay in analysis of microscopy characterization data and hence the need to address the implication of using different models (for example regression or segmentation), data augmentations, and pretrain domains on the deployment of the models for practical applications. In this study, we perform an in-depth analysis of the prediction of crystal coverage (the proportion of the substrate covered with grown crystal) in WSe2 thin film atomic force microscopy (AFM) micrographs using regression and segmentation models. As limited data is typically available for training the models, we have used image patches and transfer learning fromImageNet and MicroNet pretrain domains. Our results show segmentation models excelled in determining crystal coverage on image patches. However, when applied to full images rather than patches, the performance of segmentation models degraded considerably, while the regressors did not, suggesting that regression models may be more robust to scale and dimension changes compared to segmentation models. The results demonstrate the efficacy of computer vision models for automating sample characterization in 2D materials while providing important practical considerations for their use in the development of chalcogenide thin films.
Publication: Moses, I. A., Wu, C., & Reinhart, W. F. (2024). Materials Today Advances,22, 100483.
Publication: 10.1016/j.mtadv.2024.100483; Data: 10.26207/g8tv-fv21
2DCC Role:The MOCVD facility was used to grow the WSe2 thin films used in this project. Machine learning was applied to 2DCC WSe2 characterization datasets via API access to the Lifetime Sample Tracking (LiST) tool .