Multi-modal Dataset of Process Parameters and Defect Images in Laser Cladding

To address the absence of a publicly available standard dataset for laser cladding defects in current academia, this study introduces a novel Multi-modal Laser Cladding Defect Dataset. Designed to reflect authentic industrial operating conditions, the dataset compiles a substantial collection of aligned multi-modal "image-parameter" samples sourced from existing literature. It covers distinct processing modes—continuous wave and pulsed laser—and records typical experimental data across a broad spectrum of core parameters, namely laser power, scanning speed, and powder feeding rate. Strictly adhering to the defect types defined in the literature, the dataset is evenly balanced into explicit categories: normal (defect-free), crack, and pore, with each class containing an equal volume of independent data subsets. This constitutes a comprehensive resource spanning diverse process scenarios.