Related Experiment Video
Updated: Jun 14, 2025

Author Spotlight: Efficient Image Recognition Using Directional Gradient Histogram Technique and Support Vector Machines
Published on: January 5, 2024
Concrete Surface Crack Detection Algorithm Based on Improved YOLOv8
Xuwei Dong1, Yang Liu1, Jinpeng Dai2
1Key Laboratory of Opto-Electronic Technology and Intelligent Control, Ministry of Education, Lanzhou Jiaotong University, Lanzhou 730070, China.
This study introduces YOLOv8-CD, an improved lightweight algorithm for concrete crack detection. It enhances accuracy and efficiency for infrastructure safety, offering real-time monitoring capabilities.
Area of Science:
- Civil Engineering
- Computer Vision
- Artificial Intelligence
Background:
- Concrete infrastructure safety relies on accurate crack detection for timely repairs.
- Existing methods face challenges like high costs, slow processing, and limited mobile application.
- Developing efficient and accurate automated crack detection is crucial for structural health monitoring.
Purpose of the Study:
- To propose an improved lightweight algorithm, YOLOv8-Crack Detection (YOLOv8-CD), for enhanced concrete surface crack detection.
- To address limitations of existing methods by improving accuracy, speed, and computational efficiency.
- To adapt advanced deep learning modules for effective extraction of crack features.
Main Methods:
- An improved YOLOv8 architecture (YOLOv8-CD) was developed, integrating Visual Attention Networks (VANs) and Large Convolutional Attention (LCA) modules.
- A novel Large Separable Kernel Attention (LSKA) module was introduced to extract crack-specific features.
- Optimizations included using the Ghost module in the backbone and GSConv with VoV-GSCSP in the neck for reduced computational complexity.
Main Results:
- The YOLOv8-CD algorithm demonstrated significant improvements in mean Average Precision (mAP50 and mAP50-95) on RDD2022 and Wall Crack datasets.
- Achieved a reduced computational load (7.9 × 10^9) and a detection speed of 88 FPS for real-time performance.
- Outperformed other mainstream object detection algorithms in concrete crack detection tasks.
Conclusions:
- The proposed YOLOv8-CD algorithm offers a superior, lightweight, and efficient solution for concrete surface crack detection.
- The integration of LSKA, Ghost module, and GSConv effectively enhances feature extraction and reduces computational cost.
- This approach enables real-time, accurate monitoring crucial for infrastructure safety and maintenance.
Related Concept Videos
Microcracking in Concrete
Non-destructive Tests for Concrete Strength
Types of Non-structural Cracks in Concrete
Plastic shrinkage cracks typically form within hours after the concrete is poured. The concrete's surface dries faster than the bottom, creating tensile stress that the still-plastic concrete cannot withstand, leading to diagonal or randomly patterned cracks on the concrete surface.
Shrinkage in Concrete
When concrete is still in its plastic state, it can undergo a decrease in volume by about 1% of its absolute volume. This decrease is known as plastic shrinkage. It arises either...
Elasticity in Concrete
Design Example: Joints in Concrete Pavements
Contraction joints are typically formed by sawing a groove into the concrete shortly after it has hardened. This creates a weakened vertical plane, deliberately encouraging cracking at...

