TY - GEN
T1 - Computer-Vision-Based Real-Time Rock Fragment Recognition During Tunnel Excavation
AU - Xu, Yang
AU - Li, Hui
AU - Qiao, Weidong
N1 - Publisher Copyright:
© IABSE Congress Nanjing 2022 - Bridges and Structures: Connection, Integration and Harmonisation, Report. All rights reserved.
PY - 2022
Y1 - 2022
N2 - Timely recognition of rock fragments can help predict the deformation of the tunnel during tunnel boring machine (TBM) tunneling. Traditional manual inspection highly relies on subjective judgments of operators and conducting sieving tests is not real-time. Rock fragments in the real-world are often observed against a dark background, distributed with high size diversity, complicatedly distributed, and blocked by each other. This study proposes a computer vision-based method for on-site rock fragments recognition. The proposed method consists of an image preprocessing module, an instance segmentation model, and a post-processing module. The results show that the pixel-level rock fragment recognition takes 0.15s for processing a 512×512 patch on average and 88% of rock fragments can be recognized. The predicted size distributions of the major and minor axis lengths of the rock fragments fit well with the ground-truth ones statistically.
AB - Timely recognition of rock fragments can help predict the deformation of the tunnel during tunnel boring machine (TBM) tunneling. Traditional manual inspection highly relies on subjective judgments of operators and conducting sieving tests is not real-time. Rock fragments in the real-world are often observed against a dark background, distributed with high size diversity, complicatedly distributed, and blocked by each other. This study proposes a computer vision-based method for on-site rock fragments recognition. The proposed method consists of an image preprocessing module, an instance segmentation model, and a post-processing module. The results show that the pixel-level rock fragment recognition takes 0.15s for processing a 512×512 patch on average and 88% of rock fragments can be recognized. The predicted size distributions of the major and minor axis lengths of the rock fragments fit well with the ground-truth ones statistically.
KW - computer vision
KW - instance segmentation
KW - rock fragment recognition
KW - tunnel boring machine tunnelling
UR - https://www.scopus.com/pages/publications/85142869390
M3 - 会议稿件
AN - SCOPUS:85142869390
T3 - IABSE Congress Nanjing 2022 - Bridges and Structures: Connection, Integration and Harmonisation, Report
SP - 1240
EP - 1247
BT - IABSE Congress Nanjing 2022 - Bridges and Structures
PB - International Association for Bridge and Structural Engineering (IABSE)
T2 - IABSE Congress Nanjing 2022 - Bridges and Structures: Connection, Integration and Harmonisation
Y2 - 21 September 2022 through 23 September 2022
ER -