Granulometry transformer: image-based granulometry of concrete aggregate for an automated concrete production control

Abstract: The size distribution of concrete aggregate significantly affects the quality characteristics of the final concrete. However, despite its large influence, only approximate knowledge about the size distribution is known during concrete mix design and production, leading to a diminished ability of controlling target concrete properties. To overcome this limitation and to allow a precise control of desired concrete properties, we present `Granulometry Transformer’, a vision-based approach for an automated aggregate grading curve estimation from image data. Our approach demonstrates state-of-the-art results on two challenging public benchmark data sets of both, coarse and fine aggregate material.