Traffic pedestrian detection algorithm based on lightweight SSD

JiaBao Huang; Qiong Cai; Yu Chen; QianQian Huang; Fang Li

doi:10.1117/12.2628565

7 March 2022 Traffic pedestrian detection algorithm based on lightweight SSD

JiaBao Huang, Qiong Cai, Yu Chen, QianQian Huang, Fang Li

Proceedings Volume 12167, Third International Conference on Electronics and Communication; Network and Computer Technology (ECNCT 2021); 121672C (2022) https://doi.org/10.1117/12.2628565
Event: 2021 Third International Conference on Electronics and Communication, Network and Computer Technology, 2021, Harbin, China

Abstract

With the development of urbanization, pedestrian detection has become an important link in road traffic. Some existing pedestrian detection algorithm models are too large and the detection speed is too slow, while the accuracy of lightweight pedestrian detection can’t meet the detection requirements. Therefore, a detection algorithm that can meet the accuracy and real-time is designed. In this paper, SSD is selected as the basic model, MobileNetV2 is used as the backbone network, and deconvolution multi-scale feature fusion is added to the backbone network, ECA-Net high-efficiency channel attention module is added to the feature extraction network, and HFF multi-hierarchy structure is added to the auxiliary network to improve the detection ability of the model. In addition to ensuring the lightweight of the model, the inspection effect of the model can be improved. The experimental results show that the model designed in this paper obtains 78.6% mAP on Caltech dataset, the model size is 28M. Compared to the original model and other lightweight SSD models with higher accuracy, faster and smaller models.

Citation Download Citation

JiaBao Huang, Qiong Cai, Yu Chen, QianQian Huang, and Fang Li "Traffic pedestrian detection algorithm based on lightweight SSD", Proc. SPIE 12167, Third International Conference on Electronics and Communication; Network and Computer Technology (ECNCT 2021), 121672C (7 March 2022); https://doi.org/10.1117/12.2628565

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