Abstract
This dataset comprises 593 high-resolution RGB images of machined workpiece surfaces, acquired using an industrial borescope integrated with a microscope camera and adjustable white LED illumination. Images are captured at a resolution of 2592 × 1944 pixels and 300 pixels per inch, providing detailed visualization of both inner and outer surfaces. The dataset is organized into four distinct workpiece categories (Piece01 to Piece04), with each image manually annotated by experts to indicate the presence or absence of surface defects, including wear consistent with quality standards. Images are systematically arranged within a hierarchical directory structure, facilitating straightforward access by workpiece and defect status. The acquisition system was designed to ensure controlled illumination and imaging conditions, enabling consistent capture of surface detail critical for automated defect analysis. This expertly labeled dataset supports the development and benchmarking of computer vision algorithms for surface inspection in manufacturing, with a particular focus on defect detection in components made from the same material under varying wear conditions. The comprehensive coverage and structured organization promote reproducibility and facilitate tailored investigations into machining quality and automated inspection methodologies. All dataset materials are publicly accessible to enable further research and industrial application.