Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Road Surface Defect Detection -- From Image-based to Non-image-based: A Survey

Jongmin Yu, Jiaqi Jiang, Sebastiano Fichera, Paolo Paoletti, Lisa Layzell, Devansh Mehta, Shan Luo

TL;DR

This survey addresses road surface defect detection by categorizing methods and data sources from traditional RGB image approaches to depth-based and fusion strategies. It highlights the dominance of image-based techniques while documenting a growing interest in depth sensing (ToF, structured light, LiDAR) and 3D data for improved defect localization and sizing. The paper catalogues extensive RGB datasets (pixel-level and bounding-box annotations) and depth datasets, noting the lack of publicly available point-cloud-only collections. It also discusses challenges such as weather/lighting robustness, calibration demands, and the need for standardized datasets to advance non-image-based methods. Overall, the work maps current tools and datasets, guiding future development toward robust, scalable road defect detection and smarter infrastructure management.

Abstract

Ensuring traffic safety is crucial, which necessitates the detection and prevention of road surface defects. As a result, there has been a growing interest in the literature on the subject, leading to the development of various road surface defect detection methods. The methods for detecting road defects can be categorised in various ways depending on the input data types or training methodologies. The predominant approach involves image-based methods, which analyse pixel intensities and surface textures to identify defects. Despite their popularity, image-based methods share the distinct limitation of vulnerability to weather and lighting changes. To address this issue, researchers have explored the use of additional sensors, such as laser scanners or LiDARs, providing explicit depth information to enable the detection of defects in terms of scale and volume. However, the exploration of data beyond images has not been sufficiently investigated. In this survey paper, we provide a comprehensive review of road surface defect detection studies, categorising them based on input data types and methodologies used. Additionally, we review recently proposed non-image-based methods and discuss several challenges and open problems associated with these techniques.

Road Surface Defect Detection -- From Image-based to Non-image-based: A Survey

TL;DR

This survey addresses road surface defect detection by categorizing methods and data sources from traditional RGB image approaches to depth-based and fusion strategies. It highlights the dominance of image-based techniques while documenting a growing interest in depth sensing (ToF, structured light, LiDAR) and 3D data for improved defect localization and sizing. The paper catalogues extensive RGB datasets (pixel-level and bounding-box annotations) and depth datasets, noting the lack of publicly available point-cloud-only collections. It also discusses challenges such as weather/lighting robustness, calibration demands, and the need for standardized datasets to advance non-image-based methods. Overall, the work maps current tools and datasets, guiding future development toward robust, scalable road defect detection and smarter infrastructure management.

Abstract

Ensuring traffic safety is crucial, which necessitates the detection and prevention of road surface defects. As a result, there has been a growing interest in the literature on the subject, leading to the development of various road surface defect detection methods. The methods for detecting road defects can be categorised in various ways depending on the input data types or training methodologies. The predominant approach involves image-based methods, which analyse pixel intensities and surface textures to identify defects. Despite their popularity, image-based methods share the distinct limitation of vulnerability to weather and lighting changes. To address this issue, researchers have explored the use of additional sensors, such as laser scanners or LiDARs, providing explicit depth information to enable the detection of defects in terms of scale and volume. However, the exploration of data beyond images has not been sufficiently investigated. In this survey paper, we provide a comprehensive review of road surface defect detection studies, categorising them based on input data types and methodologies used. Additionally, we review recently proposed non-image-based methods and discuss several challenges and open problems associated with these techniques.
Paper Structure (13 sections, 3 figures, 3 tables)

This paper contains 13 sections, 3 figures, 3 tables.

Table of Contents

  1. Road types and defects
  2. Road types depending on surfacing materials
  3. Types of road defects
  4. Data collection protocols
  5. RGB Camera
  6. Stereo cameras
  7. ToF-based and Structured light-based depth sensors
  8. Road defect detection datasets
  9. RGB image-based datasets
  10. Pixel-level annotated datasets
  11. Bounding-box annotated datasets
  12. Depth sensor-based datasets
  13. Conclusion

Figures (3)

  • Figure 1: Images of various types of roads based on the materials used for road surfacing: (a) Earthen road, (b) Gravel road, (c) Kankar road, (d) WBM road, (e) Asphalt road, and (f) Concrete road.
  • Figure 2: Example snapshots of various types of road surface defects: (a) Transverse crack, (b) Longitudinal track, (c) Alligator crack, (d) Block crack, and (e) Pothole. It is important to note that T the key difference between (a) and (b) is the angle with the road lane. If the crack runs parallel to the road lane, it is classified as a transverse crack. Conversely, if the crack is at a right angle to the lane, it is classified as a longitudinal crack.
  • Figure 3: Example snapshots of data collected from various sensors, including RGB cameras, stereo cameras, and Time of Flight (ToF) based depth sensors: (a) An image from the CrackForest dataset shi2016automatic taken by an iPhone 5's mobile camera. (b) An image of a road surface and the corresponding disparity map obtained by Fan et al. fan2018real. (c) Visualisation results of point clouds collected by Zhu et al. zhu2020measurement.