Reinforced concrete structure segmentation dataset 1841
Description
This dataset is a comprehensive, pixel-level annotated image dataset for multi-type damage detection and segmentation in reinforced concrete (RC) structures. This dataset is freely available for academic and research purposes, including direct use in training, validation, benchmarking, and comparative studies of damage detection and segmentation algorithms. If you use this dataset in any publication, report, or derivative work, please cite the following related papers: Wang, J., & Ueda, T. (2025). Automatic damage detection and segmentation using deep learning algorithms in reinforced concrete structure inspections. Structural Concrete, 26(5), 5511–5534. Wang, J., Wang, Z., Wang, Y., & Li, Z. (2025). Automated multi-type damage detection framework in reinforced concrete structures via data augmentation and deep segmentation networks. Journal of Civil Structural Health Monitoring, 15(8), 3861–3884. The dataset consists of 1,841 concrete surface images. All images were manually annotated by the authors to ensure annotation consistency and high labeling accuracy. The dataset covers common damage categories encountered in RC structures: concrete cracks, rebar corrosion, and concrete spalling, each with varying severity levels primarily considering their different manifestations, causes, and development. These images are further divided into eight common damage categories: (1) non-structural cracks, (2) structural cracks, (3) minor spalling, (4) moderate spalling, (5) major spalling/delamination, (6) rebar corrosion (spalling with exposed corroded reinforcement), (7) concrete crushing, (8) structural deformation. Each damage category is encoded using a distinct color in the segmentation masks: background (gray), non-structural crack (red), structural crack (green), minor spalling (cyan), moderate spalling (purple), major spalling (orange), spalling with exposed corroded reinforcement (yellow), concrete crushing (blue), and structural deformation (magenta).