Intelligent pervious concrete pore identification and segmentation method based on deep learning

Abstract

A pervious concrete pore intelligent identification and segmentation method based on deep learning comprises the following steps: preparing a pervious concrete sample, performing CT scanning on the pervious concrete sample to obtain a slice image sample of the sample, and taking the slice image sample as an initial data set; performing standardized preprocessing, data set labeling and data set enhancement on the initial data set to obtain two groups of standard data sets; segmenting the standard data set into a training set, a verification set and a test set; constructing a Mask R-CNN deep learning model, setting hyper-parameters on the training set and performing model training on the model, continuously and automatically adjusting the hyper-parameters through a representation result on the verification set, and saving to obtain an optimal learning model; and inputting the image in the test set into the optimal learning model to carry out pore intelligent identification and segmentation, outputting to obtain a pore segmentation mask pattern, the pore number, the pore area and the porosity, and giving an evaluation index at the same time. According to the method, the precision and the efficiency are greatly improved, and the robustness and the generalization ability are higher.

Description

technical field [0001] The invention relates to a method for analyzing the pore structure of permeable concrete, in particular to a method for intelligent recognition and segmentation of pores in permeable concrete based on deep learning. Background technique [0002] Permeable concrete is a new type of paving material with various environmental benefits such as water seepage, water purification, noise reduction, and mitigation of heat island effect, and it is used more and more widely. The complex pore structure is the basic characteristic of pervious concrete, which has been proved to be the decisive factor affecting the macroscopic performance of pervious concrete. The results of pore identification and segmentation are the preconditions for analyzing and characterizing the pore structure of pervious concrete, and its accuracy will directly determine the accuracy of subsequent performance analysis. [0003] At present, there are two main methods for analyzing the pore st...

AI Technical Summary

Problems solved by technology

[0004] In order to solve the problem of low efficiency and low accuracy, the artificial threshold method in the image processing technology used in the current permeable concrete pore recognition and segmentation requires manual operation

Images