.png){kind=icon}
.png){kind=icon}
Return
Improved U-net network-based segmentation method for fast diffusion measurement
.png){kind=icon}
PRE
AI
DOI：10.1016/j.optlaseng.2024.108741.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
The diffusion of concrete is an important index to measure the fluidity of concrete, which directly affects the performance and construction quality of concrete. The current measurement methods mainly rely on manual, which leads to large errors and slow times. To solve these issues, a diffusion measurement system of fluid concrete based on an improved U-Net network is proposed to tackle these problems. This system realizes the fully automatic measurement of the diffusion degree of concrete. Firstly, an improved U-Net network is proposed for semantic segmentation of concrete images to obtain accurate object boundary information. Then, the pixel contour is extracted by edge detection, and the diffusion degree of concrete is accurately measured by the coordinate conversion principle. In addition, different semantic segmentation models and diffusion calculation algorithms are compared. The comparison results show that the improved U-Net parameter has been reduced to 1.88 M, and the mIoU (mean intersection over union) has been increased from 95.25% to 98.74%. Furthermore, the experiment tests show that the proposed method can automatically and accurately measure the diffusion of concrete with an average error of about 3mm.
Keywords:
Diffusion
Semantic segmentation
Edge detection
Concrete
Journal
IF:
3.7
Papers:
7.1K
Citations:
1.7W