Simplified GoogLeNet-based spiking neural network for bridge crack detection

Ziqi Lin; Wujian Ye

doi:10.1117/12.2684977

19 October 2023 Simplified GoogLeNet-based spiking neural network for bridge crack detection

Ziqi Lin, Wujian Ye

Proceedings Volume 12709, Fourth International Conference on Artificial Intelligence and Electromechanical Automation (AIEA 2023); 1270945 (2023) https://doi.org/10.1117/12.2684977
Event: Fourth International Conference on Artificial Intelligence and Electromechanical Automation (AIEA 2023), 2023, Nanjing, China

Abstract

Concrete bridge cracking is a common problem in practical engineering. To address the problems of many model parameters and long detection time in current deep learning methods, a simplified GoogLeNet model based on spiking neural network to reduce the overhead of using inspection equipment in engineering and improve the efficiency of detection in engineering. First, a simplified GoogLeNet model is proposed by analyzing the advantages of the original GoogLeNet network model. Then, a simplified GoogLeNet model based on the Spiking neural network is built and the network is trained by the surrogate gradient. Finally, we use concrete bridge crack dataset for validation and comparison with several mainstream detection methods. The experimental results show that the accuracy of the network model proposed in this paper is close to that of the mainstream detection methods, and the size of the model proposed in this paper is reduced by 89% and the inference time is reduced by 22% compared with the VGG7 model based on spiking neural network.

Citation Download Citation

Ziqi Lin and Wujian Ye "Simplified GoogLeNet-based spiking neural network for bridge crack detection", Proc. SPIE 12709, Fourth International Conference on Artificial Intelligence and Electromechanical Automation (AIEA 2023), 1270945 (19 October 2023); https://doi.org/10.1117/12.2684977

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