Parallel Structural Analysis Using Threads

593

336

This article implements the state‐of‐the‐art deep learning technologies for a civil engineering application, namely recognition of structural damage from images. Inspired by ImageNet Challenge and the development of computer hardware, the concept of Structural ImageNet is proposed herein with four naïve baseline recognition tasks: component type identification, spalling condition check, damage level evaluation, and damage type determination. A relatively small number of images (2,000) are selected from the Structural ImageNet and manually labeled according to the four recognition tasks. In order to avoid overfitting, Transfer Learning (TL) based on VGGNet (Visual Geometry Group) is introduced and applied using two different strategies, namely feature extractor and fine‐tuning. Two experiments are designed based on properties of these two strategies to find the relative optimal model parameters and scope of application. Models obtained by both strategies indicate the promising recognition results and different application potentials where feature extractor and fine‐tuning can be respectively used for preliminary analysis and for further improvement. These results also reveal the potential uses of deep TL in image‐based structural damage recognition.

Section: Transfer Learningmentioning

confidence: 99%

Deep Transfer Learning for Image‐Based Structural Damage Recognition

Gao

Mosalam

2018

593

336

“…() failed to identify many of the cracks due to the intensity inhomogeneity as summarized in Table . In addition, further improvement on the implementation of the proposed method using large‐scale parallelized computation techniques and capacities (Adeli and Kamal, , ,b; Adeli and Vishnubhotla, ) should dramatically improve the processing speed. Overall, the proposed method has shown promising results in accurately and efficiently detecting the cracks in the nonballasted rail track slabs.…”

Section: Experimental Testsmentioning

confidence: 99%

A Nonballasted Rail Track Slab Crack Identification Method Using a Level‐Set‐Based Active Contour Model

Qiu

et al. 2018

The nonballasted rail tracks have been extensively applied in the new high‐speed railway system development in China, that is, the China Railway Track System (CRTS). However, many defects have been identified during the operation of the CRTS, among which concrete slab crack is recognized as one of the most common, yet critical defects that require accurate identification and timely maintenance attention. Because of the unique outlook of the cracks in nonballasted rail track slabs captured in the survey imagery, direct adaptations of the existing crack extraction methods show dramatically degraded performance. A new automatic crack identification method is developed in this study by employing a region‐based active contour framework with the intensity cluster energy. The proposed method embodies three major contributions, including (1) a heavy penalization energy component that could effectively avoid both under‐ and overevolutions; (2) a multiphase level‐set function that effectively evolves the contours generated with different intensity clusters; and (3) a two‐step implementation of the framework that significantly improves the efficiency. The experimental test evaluates the performance of the proposed method using the data collected on high‐speed rail tracks in Hebei Province, China. The proposed method accurately identifies more than 93.0% of digitized cracks in different crack patterns and challenging backgrounds using a data set consisting of 1,500 synthetic images and 150 actual images. In addition, it shows promising performance in comparison with other popular state‐of‐the‐art crack detection algorithms in terms of accuracy and computational efficiency. The proposed method has demonstrated the promising capacity to support a reliable and efficient nonballasted rail track crack identification and to facilitate the subsequent maintenance cost‐effectively.

“…Although the real‐time application is not within the scope of this study for crack width measurement, it is a desirable feature for large‐scale pavement crack evaluation. In the future implementation, optimization of the current algorithms and parallelization of the proposed method are recommended to take full advantage of the parallel computing (Tsai and Huang, ; Torbol, ) and large‐scale engineering computation capacities in pavement and structure engineering (Adeli and Vishnubhotla, ; Adeli and Kamal, ; Adeli and Kamal, ; Adeli and Kamal, ).…”

Section: Experimental Testsmentioning

confidence: 99%

Pavement Crack Width Measurement Based on Laplace's Equation for Continuity and Unambiguity

Wang

Zhang

Wang

et al. 2017

Crack is one of the most important pavement condition indicators that are immediately relevant to water ingress and pavement deterioration. In practices of pavement management, crack width has been extensively referenced by highway agencies to determine pavement crack severity. Accurate measurement of pavement crack width is meaningful for highway agencies in understanding the mechanism of crack formation, and in predicting crack propagation. This article presents a new automatic method for measuring crack width using the binary crack map images. The proposed method introduces a new crack width definition and formulates it using the Laplace's Equation so that crack width can be continuously and unambiguously measured. Two algorithms, including the crack blob extraction algorithm and the crack