A Systematic Review of Transformer Architecture for Infrastructure Evaluation and Monitoring

Abstract: Abstract:
Highways and railroads are foundational elements of transportation networks, playing a crucial role in fostering economic growth and development. Ensuring the safety and efficiency of these systems is vital, necessitating the integration of cost-effective and autonomous monitoring technologies. In recent years, advancements in deep learning, particularly with the large language transformer models, have spurred innovative monitoring approaches that enable reliable and efficient infrastructure condition assessment. This paper presents a comprehensive review of transformer based methodologies specifically developed for the automated evaluation of pavements and railway assets. These approaches have demonstrated significant capabilities in detecting and segmenting pavement cracks, identifying pavement surface distresses, and uncovering faults within the various components of rail structures. By summarizing the latest research findings and practical applications in the field, this review underscores the potential of transformer architectures to significantly contribute to the maintenance and enhancement of transportation infrastructure resilience. The findings presented in this review not only offer insights into the current practices but also emphasize the transformative role transformer models can play in advancing infrastructure monitoring technologies, ultimately ensuring safer and more durable transportation networks that can adapt to future demands and challenges.

Key words: Transformer Architecture, Automated Monitoring, Pavement Evaluation, Railway Infrastructure