A comprehensive survey on multimodal computer vision (CV)-assisted corrosion assessment and corrosion prediction

Corrosion assessment and lifespan prediction, as core technologies in the lifecycle management of industrial metallic equipment, play an irreplaceable role in ensuring operational safety, optimising maintenance strategies and advancing industrial sustainability. With the rise of the fourth scientific paradigm of data-driven approaches, significant challenges have emerged to the third scientific paradigm of mathematical model-driven methods in the field of metal corrosion and protection. Corrosion image-based evaluation stands as one of the most intuitive methods among existing techniques. However, traditional image processing approaches for material surface corrosion suffer from limitations such as high dependency on image quality, inability to detect early-stage corrosion and restricted focus on surface-level information. These challenges have catalysed the emergence of the fifth paradigm, AI for Science (AI4S), in this domain. This review innovatively proposes a multimodal computer vision (CV)-assisted intelligent corrosion assessment framework. By systematically analysing advancements in CV for core tasks – including decoupling corrosion morphology features, cross-modal damage correlation modelling and corrosion dynamics evolution prediction – this work establishes a tripartite intelligent corrosion analysis framework integrating ‘multi-source data perception, feature synergy interpretation and mechanism-driven prediction’. It explores how CV enhances corrosion detection accuracy and lifespan prediction reliability through applications in corrosion classification, detection, segmentation, prediction and mechanistic analysis. The proposed multimodal fusion technology, coupled with deep learning-based intelligent corrosion type recognition and lifespan prediction models, represents critical future directions. These advancements hold significant potential for advancing intelligent corrosion monitoring and reducing industrial maintenance costs.