Deep learning-based particle gradation detection of fillers

Deep learning is one of the key topics of research today. Researchers have found that deep learning algorithms have developed rapidly in recent years, and this paper aims to apply deep learning algorithms to practical problems in civil engineering. Therefore, in this paper, we take images of particles on a construction site and segment the images using the YOLO model, and then calculate the grading curves of the particles using the grading detection model. It is found that the use of deep learning to calculate the grading curve of particles is rapid and convenient compared with the traditional experimental methods; the accuracy is slightly flawed compared with the traditional experimental methods. With the advancement of algorithms and hardware upgrading, the accuracy of this method can be comparable to the traditional experimental method, and eventually, it is widely adopted by the site.