Deep learning-based pixel-level rock fragment recognition during tunnel excavation using instance segmentation model

Qiao, Weidong; Zhao, Yufei; Xu, Yang; Lei, Yu-Meng; Wang, Yujie; Yu, Shu; Li, Hui

doi:10.1016/j.tust.2021.104072

Cited by 34 publications

(10 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…These cases exhibit noticeable and clear dynamic blur, respectively. Additionally, several comparable methods were chosen for comparison in the experiment, including CBF (23), VSMWLS (24), Densefuse-Add (25), Densefuse-L1 (25), NCVE (26), and CNN (27). After comparing the fusion results of several different networks, the impact on the quality of the image is visually demonstrated.…”

Section: Resultsmentioning

confidence: 99%

“…Zhao et al [22] developed a deep learningbased approach to extend the PANet model by adding semantic branches to reduce inaccuracies associated with crack discontinuities and image skeletonization. Qiao et al [23] designed a target detection subnetwork based on an improved single-time detector architecture to localize the defective area using multilevel feature fusion, a priori anchors, and self-attention modules. Zhao et al [24] proposed an approach for integrated processing and scaling of images for shield tunnel lining leakage region classification and instance segmentation.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Multi visual images fusion approach for metro tunnel defects based on saliency optimization of pixel level defect image features

Qiu,

Zhu,

Wang

et al. 2024

Meas. Sci. Technol.

View full text Add to dashboard Cite

The multi vision metro tunnel defect sensing system mainly consists of IRT and RGB cameras, which can automatically identify and extract small tunnel lining surface defects, greatly improving detection efficiency. However, the presence of various issues like train vibration, inconsistent lighting, fluctuations in temperature and humidity leads to the images showing inadequate uniformity in illumination, blurriness, and a decrease in the level of detail. The above issues have led to unsatisfactory fusion processing results for multiple visual images and increased missed detection rates. A multi visual images fusion approach for metro tunnel defects based on saliency optimization of pixel level defect image features is proposed. This method first takes the motion state of the train and the blurry image as constraints to eliminate dynamic blurring in the image. Secondly, Image weights are allocated based on the uniformity of visible light image illumination in the tunnel, as well as real-time temperature and humidity. Finally, image feature extraction and fusion are performed by a U-Net network that integrates channel attention mechanisms. The experimental results demonstrate that this approach improves the image pixel value variation rate by 39.7%, enhances the edge quality by 23%, and outperforms similar approach in terms of average gradient, gradient quality, and sum of difference correlation with improvements of 15.9%, 7.3%, and 26.6% respectively.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Multi visual images fusion approach for metro tunnel defects based on saliency optimization of pixel level defect image features

Qiu,

Zhu,

Wang

et al. 2024

Meas. Sci. Technol.

View full text Add to dashboard Cite

show abstract

“…Guojian et al (2021) trained a large number of rock slice images through the SqueezeNet network model, which not only ensured good classification accuracy but also greatly reduced model parameters and improved the classification speed, which lays the foundation for realizing the movable rock classification system. Qiao et al (2021) applied a deep‐learning‐based method for pixel‐level rock fragment recognition during tunnel excavation. Similarly, Chen et al (2022) adopted a deep‐learning‐based model for the rock mass condition assessment during tunnel boring machine (TBM) tunnel excavation.…”

Section: Introductionmentioning

confidence: 99%

Deep learning‐based recognition method of red bed soft rock image

et al. 2023

View full text Add to dashboard Cite

In order to improve the investigation efficiency, improve the traditional engineering investigation work mode, realize the intelligent investigation, and improve the economic and social benefits of enterprises, a red soft rock image intelligent analysis and recognition system is proposed based on deep learning methods. The intelligent recognition system includes two core algorithms: soft rock image decomposition and soft rock lithology/weathering degree recognition. The research shows that the identification model of weathering degree and the lithology identification model based on the convolutional neural network (CNN) algorithm have a good identification effect, and the identification probability of moderately weathered rock reaches 95.22% and the accuracy of lithology identification is 91.34%. With the increase in training data, the recognition effect will be further improved. The intelligent identification system has been integrated into the WeChat mini programme, App, and Web system, which can be directly applied to field geological survey operations to assist geological workers in the investigation work, and realize the intelligent identification and classification of red bed soft rock.

show abstract

“…Therefore, precise damage localization and rapid condition assessment of post-earthquake buildings in urban areas are critical for emergency responses and rescue decisions. Recently, the rapid development of machine learning and computer vision techniques profoundly promoted the evolution of earthquake engineering, integrating with remote sensing, UAVs, and robot techniques [1][2][3][4][5][6][7][8][9][10][11]. Images acquired by different platforms have unique advantages and characteristics.…”

Section: Introductionmentioning

confidence: 99%

Coarse-to-Fine Seismic Assessment Based on Computer Vision

XU,

ZHANG,

2023

Proceedings of the 14th International Workshop on Structural Health Monitoring

View full text Add to dashboard Cite

Recently, remote sensing satellites, unmanned aerial vehicles (UAVs), and smartphones have been extensively utilized in non-contact post-earthquake inspection at different scales with cutting-edge computer vision and machine learning techniques. This study establishes a computer-vision-based coarse-to-fine seismic assessment framework to localize dense buildings in urban areas, classify collapsed and noncollapsed states, recognize multi-type surface damage on structural components, and evaluate seismic performances. First, a Transformer-CNN deep learning architecture is designed for semantic segmentation of dense buildings and binary classification of collapsed states using large-scale remote sensing satellite images. It consists of a Swin Transformer encoder, multi-stage feature fusion module, and UPerNet decoder to extract global correlations and local features of dense buildings synchronously. Then, a multi-task learning strategy is proposed to simultaneously recognize multi-type structural components (column, beam, wall), seismic damage (concrete crack, spalling, and rebar exposure), and multi-level damage states (minor, moderate, major) using medium-scale UAV images. It contains a CNN-based encoder-decoder backbone with skip-connection modules and multi-head segmentation subnetworks for different tasks. The geometric consistency loss of split line, area, and curvature is further designed to refine the semantic segmentation of local details, increase boundary smoothness, and suppress inner voids. Finally, a machine learning neural network is established to quantify the seismic damage index of structural components using damage-related parameters (lengths, areas, and numbers of concrete crack, spalling, and rebar exposure) and design-related parameters (axial compression ratio, shear span ratio, and volumetric stirrup ratio) as inputs. A seismic damage indicator with an explicit bound of [0,1] can be obtained to reflect the nonlinear accumulation of seismic damage. The effectiveness and applicability under real-world post-earthquake scenarios have been validated by the 2017 Mexico City Earthquake M7.1, 2008 Beichuan Earthquake M8.0, 2010 Yushu Earthquake M6.9, 2015 Nepal Earthquake M7.8, 2016 Ecuador Earthquake M7.8, and 2016 Meinong Earthquake M6.7.

show abstract

Deep learning-based pixel-level rock fragment recognition during tunnel excavation using instance segmentation model

Cited by 34 publications

References 32 publications

Multi visual images fusion approach for metro tunnel defects based on saliency optimization of pixel level defect image features

Multi visual images fusion approach for metro tunnel defects based on saliency optimization of pixel level defect image features

Deep learning‐based recognition method of red bed soft rock image

Coarse-to-Fine Seismic Assessment Based on Computer Vision

Contact Info

Product

Resources

About