A taxonomy and roadmap for corrosion detection systems: bridging sensing, data, and deployment

Abstract Corrosion remains a critical challenge affecting the safety, reliability, and service life of infrastructure across sectors such as maritime, energy, transportation, and civil engineering. Conventional inspection methods, including non-destructive testing and visual assessment, are widely used but remain limited in scalability, objectivity, and sensitivity to early-stage degradation. Recent advances in imaging spectroscopy, particularly hyperspectral and multispectral imaging, combined with machine learning, have enabled non-contact and data-driven strategies capable of corrosion-product characterization and improved detection performance. This study presents an integrative review of corrosion detection methods and introduces a unified taxonomy. The taxonomy organizes existing methods across four interdependent dimensions: sensing modality, data representation, analytical approach, and deployment context. Together, these dimensions provide an integrated perspective capturing interactions among sensing technologies, data complexity, analytical models, and operational constraints, enabling consistent comparison across heterogeneous inspection strategies. A taxonomy-driven comparative analysis of 48 representative studies reveals fundamental trade-offs among detection sensitivity, scalability, and deployment feasibility. The analysis reveals a transition from visually driven inspection to spectral and data-driven approaches, alongside the emergence of hybrid, multi-sensor, and autonomous systems. At the same time, a persistent performance–deployment gap is identified: high-sensitivity methods, particularly hyperspectral and multimodal systems, remain largely confined to laboratory environments, whereas scalable approaches such as RGB-based inspection generally provide reduced detection sensitivity. To address this gap, a roadmap for industrial adoption is proposed, emphasizing data standardization, interpretable and robust analytics, multi-sensor integration, and scalable deployment platforms. Overall, this work provides both a conceptual foundation and a practical roadmap for advancing corrosion detection toward integrated, intelligent, and field-deployable monitoring systems.