SEM Image Dataset of Metallic Powders for Flowability and Machine Learning Studies in Additive Manufacturing

This dataset contains 3,667 backscattered scanning electron microscopy (SEM) images collected from 40 metallic powders used in additive manufacturing research. Images were acquired in 16-bit TIFF format under multiple magnification levels and are accompanied by experimentally measured powder flowability properties, including Angle of Repose (AOR), Hausner Ratio (HR), Carr Index (CI), and Hall Flow Rate. The dataset was developed to support research in powder characterisation, image analysis, computer vision, machine learning, and quality control for additive manufacturing. It has been used to investigate the influence of image representation and bit-depth conversion on morphology preservation, prediction performance, model generalisation, and explainability in deep learning workflows. The dataset includes image metadata, powder identifiers, magnification information, and associated flowability measurements.