Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

TD-RD: A Top-Down Benchmark with Real-Time Framework for Road Damage Detection

Xi Xiao, Zhengji Li, Wentao Wang, Jiacheng Xie, Houjie Lin, Swalpa Kumar Roy, Tianyang Wang, Min Xu

TL;DR

This work introduces TD-RD, the first top-down road damage dataset, and TD-YOLOV10, a real-time detector enhanced with MAPSE and VGAU to exploit top-down context. The dataset comprises 7,088 high-resolution images with 12,882 labeled damage instances across cracks, potholes, and repairs, curated from a much larger pool to ensure quality and balance. TD-YOLOV10 achieves state-of-the-art or competitive performance on TD-RD and public benchmarks (CNRDD, CRDDC’22) while maintaining real-time speeds (up to 200 FPS), validating the effectiveness of multi-scale attention and global upsampling for heterogeneous road damage patterns. The work provides both a valuable top-down viewpoint dataset and a performant, efficient detection framework that can advance automated road-condition monitoring and infrastructure maintenance applications.

Abstract

Object detection has witnessed remarkable advancements over the past decade, largely driven by breakthroughs in deep learning and the proliferation of large scale datasets. However, the domain of road damage detection remains relatively under explored, despite its critical significance for applications such as infrastructure maintenance and road safety. This paper addresses this gap by introducing a novel top down benchmark that offers a complementary perspective to existing datasets, specifically tailored for road damage detection. Our proposed Top Down Road Damage Detection Dataset (TDRD) includes three primary categories of road damage cracks, potholes, and patches captured from a top down viewpoint. The dataset consists of 7,088 high resolution images, encompassing 12,882 annotated instances of road damage. Additionally, we present a novel real time object detection framework, TDYOLOV10, designed to handle the unique challenges posed by the TDRD dataset. Comparative studies with state of the art models demonstrate competitive baseline results. By releasing TDRD, we aim to accelerate research in this crucial area. A sample of the dataset will be made publicly available upon the paper's acceptance.

TD-RD: A Top-Down Benchmark with Real-Time Framework for Road Damage Detection

TL;DR

This work introduces TD-RD, the first top-down road damage dataset, and TD-YOLOV10, a real-time detector enhanced with MAPSE and VGAU to exploit top-down context. The dataset comprises 7,088 high-resolution images with 12,882 labeled damage instances across cracks, potholes, and repairs, curated from a much larger pool to ensure quality and balance. TD-YOLOV10 achieves state-of-the-art or competitive performance on TD-RD and public benchmarks (CNRDD, CRDDC’22) while maintaining real-time speeds (up to 200 FPS), validating the effectiveness of multi-scale attention and global upsampling for heterogeneous road damage patterns. The work provides both a valuable top-down viewpoint dataset and a performant, efficient detection framework that can advance automated road-condition monitoring and infrastructure maintenance applications.

Abstract

Object detection has witnessed remarkable advancements over the past decade, largely driven by breakthroughs in deep learning and the proliferation of large scale datasets. However, the domain of road damage detection remains relatively under explored, despite its critical significance for applications such as infrastructure maintenance and road safety. This paper addresses this gap by introducing a novel top down benchmark that offers a complementary perspective to existing datasets, specifically tailored for road damage detection. Our proposed Top Down Road Damage Detection Dataset (TDRD) includes three primary categories of road damage cracks, potholes, and patches captured from a top down viewpoint. The dataset consists of 7,088 high resolution images, encompassing 12,882 annotated instances of road damage. Additionally, we present a novel real time object detection framework, TDYOLOV10, designed to handle the unique challenges posed by the TDRD dataset. Comparative studies with state of the art models demonstrate competitive baseline results. By releasing TDRD, we aim to accelerate research in this crucial area. A sample of the dataset will be made publicly available upon the paper's acceptance.
Paper Structure (19 sections, 3 equations, 2 figures, 2 tables)

This paper contains 19 sections, 3 equations, 2 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Benchmark
  3. Method
  4. MAPSE-based Feature Extraction
  5. Visual Global Attentional Upsampling
  6. Experiments
  7. Datasets
  8. Evaluation Metrics
  9. Implementation details
  10. Results
  11. Comparison with existing works
  12. Results on TD-RD Dataset
  13. Results on CNRDD Dataset
  14. Results on CRDDC’22 Dataset:
  15. Ablations
  16. ...and 4 more sections

Figures (2)

  • Figure 1: Annotated Samples from current datasets and TD-RD, which contains annotation distribution and categories.
  • Figure 2: Visualization of model architecuture, which contains the architecture(A) and workflow(B) of MAPSE module and VGAU module.