The Crack Diffusion Model: An Innovative Diffusion-Based Method for Pavement Crack Detection

Abstract

Pavement crack detection is of significant importance in ensuring road safety and smooth traffic flow. However, pavement cracks come in various shapes and forms which exhibit spatial continuity, and algorithms need to adapt to different types of cracks while preserving their continuity. To address these challenges, an innovative crack detection framework, CrackDiff, based on the generative diffusion model, is proposed. It leverages the learning capabilities of the generative diffusion model for the data distribution and latent spatial relationships of cracks across different sample timesteps and generates more accurate and continuous crack segmentation results. CrackDiff uses crack images as guidance for the diffusion model and employs a multi-task UNet architecture to predict mask and noise simultaneously at each sampling step, enhancing the robustness of generations. Compared to other models, CrackDiff generates more accurate and stable results. Through experiments on the Crack500 and DeepCrack pavement datasets, CrackDiff achieves the best performance (F1 = 0.818 and mIoU = 0.841 on Crack500, and F1 = 0.841 and mIoU = 0.862 on DeepCrack).

Urban Road Surface Condition Sensing from Crowd-Sourced Trajectories Based on the Detecting and Clustering Framework

“…However, the use of smartphones, a type of non-professional measuring device, introduces considerable noise and lower information density, making the extraction of road surface features and the integration of multi-source data challenging [12,20,25,34]. Since the neural network mode has the capability of feature encoding and detecting, these methods are widely used to detect complex features.…”

“…Zang et al [17] mounted smartphones on bicycles to collect road surface information based on recorded acceleration changes and derived road surface changes using the threshold value method. These methods often require data from supplementary sensors like cameras, accelerometers, gyroscopes, and audio sensors in addition to location information from smartphone sensors [9,12,26]. Li et al [16] proposed a road surface detection method based on the continuous wavelet transform (CWT).…”

“…Then, the CWT was applied to the accelerations, and a threshold value was used to detect abnormal changes related to the road surface. However, the use of smartphones, which are non-professional measuring devices, introduces considerable noise and lower information density, making the extraction of road surface features and the integration of multi-source data challenging [12,21,26,35]. Neural network models have the capability of feature encoding and detection, making them widely used to detect complex features.…”

“…(3) Update the old cell state C t−1 to the new cell state C t . The old state is multiplied by f t to "forget" certain information, and then new candidate values are added by multiplying i t and Ct , as denoted by Equation (12).…”

“…Therefore, sensing the surface condition to gather information like BFRSs is essential for urban transportation [8,9]. Traditional methods of road surface sensing rely on professional equipment and personnel, such as LiDAR, photogrammetry, and remote sensing [8,[10][11][12][13]. These methods are time-consuming and labor-intensive, making it challenging to keep up with the rapidly changing urban road conditions.…”