Road surface defect detection based on large convolution kernel and YOLOv5

Hao Zuo; Xiaowei Niu

doi:10.1117/12.2683845

14 June 2023 Road surface defect detection based on large convolution kernel and YOLOv5

Hao Zuo, Xiaowei Niu

Proceedings Volume 12708, 3rd International Conference on Internet of Things and Smart City (IoTSC 2023）; 127081Q (2023) https://doi.org/10.1117/12.2683845
Event: 3rd International Conference on Internet of Things and Smart City (IoTSC 2023）, 2023, Chongqing, China

Abstract

With the rapid development of China's road system, road maintenance has become an important issue facing road development, and road surface defect detection is the primary link in road maintenance. In order to address the most challenging crack detection in road surface defect detection, larger convolution kernels were used in YOLOv5, which have larger receptive fields and can obtain more crack feature information. A large convolution kernel structure was also used, in which structural re-parameterization was applied to improve the detection accuracy of the model without increasing the detection speed. Furthermore, in order to further improve the detection speed of the model, depthwise separable convolution was applied to the large convolution kernel structure at the expense of sacrificing some accuracy.

Citation Download Citation

Hao Zuo and Xiaowei Niu "Road surface defect detection based on large convolution kernel and YOLOv5", Proc. SPIE 12708, 3rd International Conference on Internet of Things and Smart City (IoTSC 2023）, 127081Q (14 June 2023); https://doi.org/10.1117/12.2683845

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available