Computer vision (CV) techniques have been increasingly adopted to the quantitative analysis of metallic microstructures, offering new opportunities for automated feature extraction and microstructural characterization from microscopy images. This review presents a focused and application-oriented overview of CV-based approaches for metallic microstructure analysis, with particular emphasis on the roles and complementarities of semantic segmentation and object detection tasks. The workflow from image preprocessing and microstructural feature identification to quantitative descriptor extraction and microstructure-property correlation is systematically discussed. Representative applications in defect quantification, mechanical property evaluation, and performance-oriented materials design are summarized, together with current limitations related to data availability, generalization, and physical interpretability. This review aims to provide practical guidance and a structured reference for researchers seeking to adopt CV techniques in materials microstructure analysis. • Reviews evolution and trade-offs of CV methods for metallic microstructures • Discusses how physical knowledge improves interpretability of data-driven models • Outlines challenges and prospects of CV-driven microstructure analysis and design
Dong et al. (Sun,) studied this question.