A vehicle classification method based on deep learning and multi-attention mechanism

Kaiyan Zhang; Xinglong Feng

doi:10.1117/12.2681386

8 June 2023 A vehicle classification method based on deep learning and multi-attention mechanism

Kaiyan Zhang, Xinglong Feng

Proceedings Volume 12707, International Conference on Image, Signal Processing, and Pattern Recognition (ISPP 2023); 127074K (2023) https://doi.org/10.1117/12.2681386
Event: International Conference on Image, Signal Processing, and Pattern Recognition (ISPP 2023), 2023, Changsha, China

Abstract

Deep learning has many important applications in autonomous vehicle classification. However, due to the similarity between different types of vehicles, the deep learning model is faced with a certain degree of challenge in classifying vehicles. If the deep learning method can be improved to improve the accuracy of vehicle classification method, it will be very helpful for the practical application of autonomous driving. In this paper, ResNet34 deep learning model is selected as the backbone network, and FcaNet and several spatial attention mechanisms are added to improve the model. We tested our proposed approach on a data set containing 10 different types of vehicles. The test results on the vehicle classification data set show that the classification accuracy of the improved scheme reaches 79.0%, which is higher than the 75.42% of the original ResNet34 model. The experimental results show that the method of adding multiple attention mechanisms to ResNet deep learning network is helpful to improve the classification accuracy of different vehicles.

Citation Download Citation

Kaiyan Zhang and Xinglong Feng "A vehicle classification method based on deep learning and multi-attention mechanism", Proc. SPIE 12707, International Conference on Image, Signal Processing, and Pattern Recognition (ISPP 2023), 127074K (8 June 2023); https://doi.org/10.1117/12.2681386

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