The Size Distribution Measurement and Shape Quality Evaluation Method of Manufactured Aggregate Material Based on Deep Learning

Abstract Manufactured aggregate is a substitute for natural aggregate particles that is formed by mechanically crushing parent rock. Its particle shape has a great impact on the working performance, mechanical performance, and durability for preparing high-performance concrete. Therefore, a particle shape quality evaluation method combining deep learning and distance transformation topology is proposed. In this method, the YOLO v4 network is used to locate the particle region, and the centroid point is recognized as the feature point of this region; then, the feature points are used for distance transformation topology to approximately divide the particles area. Based on the divided results, the pixel-level segmentation result is obtained using a local threshold algorithm. The 2–8-mm limestone manufactured aggregate in a 2 million ton (1,000 kg)/year manufactured aggregate production line is carried out to demonstrate the effectiveness of the proposed method, achieving above 90 % precision in the real manufactured aggregate quality evaluation.