Abstract for presentation (Poster or Podium) with a Paper in the Conference Proceedings
Data Sensing and Analytics
Sultan Al Shafian (he/him/his)
Graduate Research Assistant
Kennesaw State University
Marietta, GA, United States
Sultan Al Shafian (he/him/his)
Graduate Research Assistant
Kennesaw State University
Marietta, GA, United States
Da Hu
Assistant Professor
Kennesaw State University
Marietta, GA, United States
Da Hu
Assistant Professor
Kennesaw State University
Marietta, GA, United States
Da Hu
Kennesaw State University
Marietta, Georgia, United States
The structural health of tunnels is crucial for their safety and durability, and cracks are key indicators of potential vulnerabilities. However, traditional inspection methods that rely on manual visual assessments are challenging, as they are labor-intensive, inconsistent, and may overlook critical defects due to human error or environmental conditions in the tunnel. This research introduces an innovative approach to tunnel crack detection, using deep learning technology. By utilizing a customized encoder-decoder network, our method provides an automated and precise solution for identifying cracks from high-resolution images of tunnels, which can be captured by drones. Our deep learning model is capable of segmenting cracks, ensuring the detection of even subtle defects that may be missed by manual inspection. Given the unique challenges that tunnels present, such as varied lighting conditions, moisture, and diverse surface textures, our model is meticulously calibrated to remain robust and accurate. Our preliminary results have shown promising outcomes, with a mean intersection over union of 0.791, highlighting the model's proficiency. By incorporating deep learning algorithms, our goal is to establish a new procedure for tunnel inspections, aiming for heightened reliability and comprehensiveness. Ultimately, our methodology seeks to enhance the safety of tunnels by ensuring timely interventions and mitigating risks associated with structural flaws.