Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

R-CNN-based complex pavement crack identification method

A pavement crack and identification method technology, which is applied in the field of complex pavement crack identification, can solve the problems of poor robustness of the identification algorithm and high complexity of identification and measurement

Inactive Publication Date: 2021-09-14
HARBIN UNIV OF SCI & TECH
View PDF8 Cites 6 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0007] The purpose of the present invention is to solve the problem that the existing highway pavement crack recognition algorithm has high recognition measurement complexity and poor robustness of the recognition algorithm due to the influence of background changes, and proposes a complex pavement crack recognition method based on R-CNN

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • R-CNN-based complex pavement crack identification method
  • R-CNN-based complex pavement crack identification method
  • R-CNN-based complex pavement crack identification method

Examples

Experimental program
Comparison scheme
Effect test

specific Embodiment approach 1

[0041] A kind of complex pavement crack recognition method based on R-CNN of this embodiment, such as figure 1 As shown, the method is realized through the following steps:

[0042] Step 1, pavement crack image preprocessing;

[0043] The performance of road pavement crack recognition based on deep learning depends on a large number of image data sets. However, in the process of road pavement crack recognition, there are problems such as insufficient training data, high data collection costs, and high labeling costs. It is very difficult to construct. The present invention adopts the migration learning method, utilizes the public dataset ImageNet with labels that has similar characteristics to road pavement cracks to pre-train Mask R-CNN, and collects the road pavement produced on the basis of pre-training. The crack data set optimizes the parameters of Mask R-CNN so that it can be used for crack identification on highway pavement.

[0044] In this way, it solves the problem...

specific Embodiment approach 2

[0055] The difference from the first embodiment is that in the R-CNN-based complex pavement crack recognition method of this embodiment, the step of pavement crack image preprocessing described in step 1 also includes the step of image expansion, specifically, for The data set is randomly mirrored, appropriate Gaussian noise is added, and the image is randomly rotated in the vertical direction to perform image enhancement. The enhancement of the data set makes up for the problem of insufficient data set samples to a certain extent, reduces the over-fitting phenomenon of the network, and can obtain a network with stronger generalization ability and better adapt to the application scenario.

specific Embodiment approach 3

[0057] Different from the second specific embodiment, in the present embodiment, a complex pavement crack recognition method based on R-CNN, the multi-layer neural network of the pavement crack recognition model described in step 2 is represented by assuming that the input of the first layer is x l , after several layers of neural network, the output is x L , then x L With the residual module, the output can be expressed as:

[0058]

[0059] In the formula, F(x i ,W i ) is the mapping output of a residual module; Wi is the weight of the i-th convolutional layer; bi is the bias of the i-th convolutional layer, assuming that the loss function output of the neural network is l, then after backpropagation, with the following results:

[0060]

[0061] It can be concluded from the above formula that when the residual module is added for backpropagation to update the weights, as the number of network layers deepens, there will be no phenomenon of gradient dispersion and g...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention discloses an R-CNN-based complex pavement crack recognition method, and belongs to the field of image recognition. An existing highway pavement crack recognition algorithm is high in recognition metering complexity and poor in robustness due to the influence of background changes. The method comprises the following steps: performing road surface crack image preprocessing: pre-training Mask R-CNN by using a public data set ImageNet, optimizing parameters, and marking road surface image data; constructing a pavement crack recognition model; performing end-to-end training on the marked pavement picture data through a deep learning algorithm to obtain an optimal result and configuration parameters of the model, and establishing a pavement crack recognition model; detecting the pavement cracks, and detecting cracks in the pictures and the positions where the cracks appear; uploading the identified crack image file to a server through a network, and storing the path of the image. The method improves the detection precision through the feature pyramid network, and improves the recognition effect of the crack.

Description

technical field [0001] The invention relates to an R-CNN-based complex pavement crack recognition method. Background technique [0002] Due to the frequent use of roads and the influence of bad weather, cracks of different shapes and degrees will appear on the road surface. If the cracks are not treated in time at the initial stage, it will not only affect the appearance of the road, reduce the comfort of the drivers, reduce the load-bearing capacity and service life of the road, but also cause more serious cracks, resulting in greater investment in maintenance funds. Seriously, it will cause road paralysis, cause traffic accidents, and endanger people's lives. Therefore, the daily maintenance of the road is essential, and keeping the road surface in good condition is the key to reducing traffic accidents and casualties. [0003] In order to objectively reflect the damage of the road, improve the service level of the road, and ensure the safety, efficiency and convenience ...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Applications(China)
IPC IPC(8): G06K9/36G06K9/46G06K9/62G06N3/04G06N3/08
CPCG06N3/084G06N3/045G06F18/24
Inventor 王微杨微
Owner HARBIN UNIV OF SCI & TECH
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com