SpotCrack: Leveraging a Lightweight Framework for Crack Segmentation in Infrastructure.

In this work, we introduce a novel methodology, “SpotCrack” designed to precisely segment road surfaces and infrastructure cracks. We employ lightweight modules to effectively identify various types of potential cracks in the infrastructure. The effectiveness of this approach is systematically evaluated on a diverse set of crack images, thereby substantiating its capacity to enhance operational efficiency within road and infrastructure management tasks. Notably, SpotCrack achieves real-time processing at an impressive rate of 130 frames per second (FPS) with an approximate inference time of 8 milliseconds, enabling its versatile applications in various real-world scenarios. This accelerated processing capability positions SpotCrack for diverse deployments, including integration with Unmanned Aerial Vehicles (UAVs) and road maintenance fleets. Significantly, SpotCrack applicability extends to individual mobility, as its high FPS empowers its integration within private vehicles, marking a transformative step in road safety, benefiting both self-driving and conventional vehicles alike. These findings underscore SpotCrack pivotal role in delivering precise crack segmentation, driving advancements in road management practices, and reinforcing road safety endeavors.