Transportation infrastructure significantly influences the development of sustainable transportation systems and the effective management of road networks. Accurate segmentation of road infrastructure provides valuable data and a structured approach for the effective management of road deterioration. However, existing methods are limited in segmenting road deterioration data due to the lack of probabilistic assessment and the inability to effectively handle outliers. To address this gap, the Gaussian mixture segmentation (GMS) model is introduced, applying a Gaussian mixture distribution to model the variability in road deterioration. Our approach utilizes spatial line segmentation to effectively segment large-scale road networks into meaningful segments. The GMS model was applied to road deterioration data from the South West region of Western Australia, identifying and segmenting roads based on their distribution characteristics using the Jensen-Shannon (JS) divergence. To assess the performance of the GMS model, metrics such as the number of segments, coefficient of variation (CV), Caliński-Harabasz Index, and Davies-Bouldin Index were evaluated. The results demonstrate that the GMS model outperformed existing segmentation methods, achieving an increase in the average percentage of segments with a CV lower than 0.25 by 23.5% to 98.8%, along with a reduction in the average Davies-Bouldin Index by 25.2% to 64.4% and improvements in the average Caliński-Harabasz Index, which increased by 23.0% to 131.1%. This approach enhances the understanding of the spatial distribution of road deterioration, informing maintenance strategies for large-scale road networks and addressing the complexities of road infrastructure management, including traffic dynamics and environmental impacts.