Deep Learning Methods for Microstructural Image Analysis: The State-of-the-Art and Future Perspectives

Finding quantitative descriptors representing the microstructural features of a given material is an ongoing research area in the paradigm of Materials-by-Design. Historically, the microstructural analysis mostly relies on qualitative descriptions. However, to build a robust and accurate process-structure-properties relationship, which is required for designing new advanced high-performance materials, the extraction of quantitative and meaningful statistical data from microstructural analysis is a critical step. In recent years, deep learning-based computer vision (CV) methods, such as those centered around convolutional neural network, generative networks, and Transformers, have shown promising results for this purpose. In this paper, we first summarize the recent deep learning methods used in computer vision and then we survey the state-of-the-art CV methods that have been applied to various multi-scale microstructural image analysis tasks, including classification, semantic segmentation, object detection, feature extraction, reconstruction, and object tracking. Additionally, we identify the main challenges and their potential solutions in the application of these CV methods to microscopy image analysis. Lastly, we discuss possible future directions of research in this area.