Automated Road Crack Localization to Guide Highway Maintenance Journal: Journal
Steffen Knoblauch, Ram Kumar Muthusamy, Pedram Ghamisi, Alexander Zipf · 2026
Read Automated Road Crack Localization to Guide Highway Maintenance Journal: Journal by Steffen Knoblauch, Ram Kumar Muthusamy, Pedram Ghamisi, Alexander Zipf (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2601-16737.txt.
Introduction
Steffen Knoblauch, Ram Kumar Muthusamy, Pedram Ghamisi, and Alexander Zipf aim to find road cracks by machine. Highways wear out fast as heat swings grow harsh. Crews must fix cracks before they spread wide. That calls for sharp maps of where cracks sit. Their 2026 preprint tests free data for that task.
Open maps and free sky shots may flag crack sites. The team checks how far such free sources go. Good flags let crews send teams to the worst spots first. That saves funds and cuts road risk. The work links earth data with road care at scale.
Catalog record
| CREATOR | Steffen Knoblauch, Ram Kumar Muthusamy, Pedram Ghamisi, Alexander Zipf |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 30 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2601-16737
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Automated Road Crack Localization to Guide Highway Maintenance Journal: Journal Steffen Knoblauch, Ram Kumar Muthusamy, Pedram Ghamisi, Alexander Zipf Abstract Highway networks are crucial for economic prosperity. Climate change-induced temperature fluctuations are exacerbating stress on road pavements, resulting in elevated maintenance costs. This underscores the need for targeted and efficient maintenance strategies. This study investigates the potential of open-source data to guide highway infrastructure maintenance. The proposed framework integrates airborne imagery and OpenStreetMap (OSM) to fine-tune YOLOv11 for highway crack localization. To demonstrate the framework’s real-world applicability, a Swiss Relative Highway Crack Density (RHCD) index was calculated to inform nationwide highway maintenance. The crack classification model achieved an F1-score of 0.84 for the positive class (crack) and 0.97 for the negative class (no crack). The Swiss RHCD index exhibited weak correlations with Long-term Land Surface Temperature Amplitudes (LT-LST-A) (Pearson’s r = − 0.05 r=-0.05 ) and Traffic Volume (TV) (Pearson’s r r = 0.17), underlining the added value of this novel index for guiding maintenance over other data. Significantly high RHCD values were observed near urban centers and intersections, providing contextual validation for the predictions. These findings highlight the value of open-source data sharing to drive innovation, ultimately enabling more efficient solutions in the public sector. 1 Introduction Critical infrastructure forms the backbone of modern society, encompassing essential systems and assets such as transportation networks (e.g., roads, bridges, and railways), energy grids (e.g., electricity and gas supply), water distribution systems, and communication networks. These infrastructures are fundamental to economic stability, public safety, and national security, as their disruption can lead to cascading failures that affect multiple sectors [ 10 ] . Among these, road infrastructure is particularly crucial, providing the foundation for mobility, facilitating the transportation of goods and people, and supporting economic activities [ 1 , 40 ] . Highways, as integral components of the road network, enable high-capacity, long-distance travel and serve as key corridors connecting major urban centers, industrial zones, and logistic hubs [ 28 ] . Their efficiency and reliability are essential for national and regional connectivity, making them a critical focus for infrastructure planning and resilience strategies [ 41 ] . To ensure that highways serve their purpose effectively, they must be maintained regularly, as external factors such as high TV, temperature fluctuations, and heavy axle loads impose significant stress on the pavement structure [ 13 , 34 ] . Without proper maintenance, these factors can lead to deterioration, reducing the safety and efficiency of highway networks, and requiring costly repairs [ 29 ] . Effective highway maintenance is essential for preserving infrastructure quality and ensuring road safety. Traditionally, maintenance decisions have been guided by reactive approaches such as damage reports, traffic volume analysis, and accident records [ 6 ] . While these methods provide valuable insights, they are inherently limited by their dependence on manual inspections and retrospective data, which may result in delayed interventions and increased maintenance costs. As a result, research has increasingly focused on leveraging advanced computational techniques to improve the accuracy and efficiency of road condition assessment [ 21 , 32 ] . Recent advancements in remote sensing and computer vision have introduced automated methods for monitoring pavement conditions. Several benchmark datasets, such as DeepCrack [ 23 ] and RDD [ 2 , 3 ] , have accelerated the development and evaluation of deep-learning models for crack detection and classification. Numerous studies have explored diverse model architectures and distress types. For instance, 24 systematically compared backbone networks such as CSPDarknet53, Hourglass-104, and EfficientNet, contributing to a better understanding of trade-offs in model performance. Broader distress categorization—covering longitudinal, transverse, and alligator cracks, potholes, delamination, and repairs—has been addressed in multi-label detection frameworks [ 47 , 50 , 33 ] . 50 introduced PDDNet, a multitask fusion model that integrates region-level and pixel-level detection, while 19 extended detection to include physical crack measurements using a modified tubularity flow field (TuFF) algorithm and distance transform method (DTM). Several studies aim to overcome data scarcity or improve robustness. 49 applied GAN-based augmentation to better train models on grooved concrete pavement imagery. 17 evaluated model stability under varying environmental conditions, including lighting and weather, which is critical for real-world deployment. Lightweight YOLO-based models - such as DenseSPH-YOLOv5 [ 31 ] , YOLOv5-CBoT [ 45 ] , YOLO-LRDD [ 42 ] , and YOLOv8-PD [ 46 ] - have been proposed for real-time applications, particularly those suited for mobile or edge devices. Despite these advancements, large-scale operational deployment remains a key bottleneck. Most existing models are tested on small-scale settings or lab-controlled benchmark datasets. 48 is among the few to attempt a scaled application, using UAV imagery and a CrackNet-based approach over 108 km of highways. 5 explored satellite imagery for crack detection, offering a broader spatial perspective but with lower resolution trade-offs. These efforts highlight the potential for scalable monitoring, yet the large-scale deployment of such methods in operational highway maintenance remains limited, with public sector practices still largely reliant on manual reporting and visual inspections. This study makes a substantive contribution toward closing the gap between academic research and practical implementation in highway maintenance. Leveraging openly available high-resolution airborne imagery and pretrained convolutional neural networks - specifically YOLOv11 - it presents an automated, scalable framework for precise crack localization on highways, capable of being routinely re-executed as new imagery becomes available. A key innovation is the Road Highway Crack Detection (RHCD) index, a novel metric designed to quantify pavement distress using remote sensing data. The framework is evaluated through a case study in Switzerland, addressing two central research questions: (RQ1) To what extent can pavement cracks on highways be detected at scale using airborne imagery? and (RQ2) To what extent do long-term land surface temperature anomalies (LT-LST-A) and traffic volume (TV) correlate with the RHCD index? The proposed approach is both flexible and transferable, offering the potential for broader application across diverse regional contexts. These contributions support the development of data-driven maintenance strategies in the public sector, advancing the transition toward a more proactive and resilient model of infrastructure management. 2 Materials and Methods This study introduces a scalable and data-driven framework for highway crack localization, aimed at supporting road infrastructure maintenance strategies at a national level. As illustrated in Figure 1 , the proposed approach comprises three main components: (1) acquisition and preprocessing of geospatial imagery - here, from Switzerland - where orthophotos are (i) retrieved via an API using OpenStreetMap highway vectors excluding tunnels, (ii) masked to exclude non-road surfaces, and (iii) filtered to remove segments obscured by overpasses such as railway bridges; (2) development of a crack localization model trained on a spatially stratified sample of image tiles, involving annotation, data augmentation, transfer learning, and performance evaluation, followed by large-scale inference across the entire highway network; and (3) aggregation of model predictions into a Relative Highway Crack Density (RHCD) index, which is subsequently analyzed in relation to TV and LT-LST-A. Figure 1 : Schematic representation of the proposed framework for scalable highway crack detection, comprising i) the retrieval of road imagery, ii) the training of a crack localization model, and iii) the generation of a RHCD index. 2.1 Data This study draws on four openly accessible datasets (see Table 1 ).