Enhancing Sidewalk Maintenance through Accurate Joint Deflection Measurement

In this work, we propose a novel real-time method for precise measurement of sidewalk joint deflections utilizing stereo vision and deep learning. This method is designed to address key challenges in infrastructure management within the context of smart cities. Our approach leverages stereo vision to establish a highly accurate Pixel-to-Meter ratio, enabling the precise conversion of image data into real-world distances. This allows for precise measurement of joint deflections by capturing depth information between adjacent sidewalk slabs. We then integrate instance segmentation, which enables the precise identification of individual sidewalk slabs, facilitating the accurate detection of joint deflections between these sections. Together, these techniques provide a robust method for detecting and measuring sidewalk joint deflections with high precision. Our results demonstrate the potential of our method for large-scale deployment in smart cities, with future work focused on integrating the system into predictive maintenance frameworks and expanding its application across diverse urban settings.