Author(s): Jaehyun Shin; Kyuhyun Park; Seongwook Choi; Dong Sop Rhee

Linked Author(s): Jaehyun Shin, Dong Sop Rhee, Seongwook Choi, KYU-HYUN PARK

Keywords: Deep learning crack UNET segmentation convolution image processing

Abstract: In order to apply damage detection methods through deep learning of crack images, an image processing algorithm that can detect crack damage from image data was utilized. Using the UNET model based on convolution among deep learning techniques, this research was able to detect cracked areas in images with classification for each pixel. Through learning and feature extraction of cracks in images through the UNET network, this semantic segmentation method showed acceptable accuracy and precision which can be applied to hydraulic structures

DOI:

Year: 2025