Preprocessed Industrial Yarn-Breakage Image Dataset (Textile Manufacturing, 2024)

This dataset provides the PREPROCESSED version of an industrial yarn-breakage image dataset acquired in a real textile manufacturing environment in 2024. The images were captured using a high-resolution industrial monochrome (Mono8) camera (2592 × 2048 pixels) mounted on a mobile inspection platform and illuminated using a controlled LED ring light.The PREPROCESSED dataset contains the same 3833 images and the same seven classes as the RAW dataset, following a strict one-to-one correspondence:000g – 71 images001g – 134 images011g – 300 images100g – 120 images101g – 416 images110g – 404 images111g – 2388 imagesA uniform preprocessing pipeline was applied to all images to improve visual clarity and robustness for machine-learning tasks. Operations include: Edge enhancement, Brightness normalization, and Gamma correction.No resizing, cropping, geometric transformations, or data augmentation was performed. All preprocessed images preserve the original resolution of 2592 × 2048 pixels (Mono8 grayscale).Two image formats are provided:BMP: Preprocessed images in BMP format.PNG: Lossless PNG versions converted from the BMP images for interoperability and convenience.This PREPROCESSED version is designed to support research in computer vision, machine learning, industrial inspection, and automated yarn-breakage detection, offering enhanced image quality while maintaining fidelity to the real-world acquisition conditions.