Open AccessA semantic segmentation method for exposed rebar on dam concrete based on Unet
Author(s) -
Jun Zhang,
Mo Jian,
Huajun Xu,
Zixing Liu
Publication year - 2020
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1651/1/012169
Subject(s) - rebar , segmentation , computer science , structural engineering , set (abstract data type) , forensic engineering , reliability engineering , artificial intelligence , engineering , programming language
Exposed rebar is an essential factor affecting the safety of the dam. In the past, manual inspection is a significant way to monitor exposed rebar risk. However, it is time-consuming, inefficient and difficult to quantitative evaluate, such as the exposed rebar area. A semantic segmentation method based on the Unet is proposed to replace the manual inspection for the dam exposed rebar automatic detection. Thirty-eight high-resolution images of dam exposed rebar are collected. Unet and the VGG16 backbone are adopted. The results indicated that Unet’s mIoU on the test set reaches 0.94, which proves to be an efficient way to detect the dam exposed rebar.