Improved semantic segmentation algorithm based on attention mechanism and pyramid module

Pengyu Zhang; Hongwei Li; Yashuai Zhao; Pengfei Zhang

doi:10.1117/12.2641143

3 October 2022 Improved semantic segmentation algorithm based on attention mechanism and pyramid module

Pengyu Zhang, Hongwei Li, Yashuai Zhao, Pengfei Zhang

Proceedings Volume 12290, International Conference on Computer Network Security and Software Engineering (CNSSE 2022); 122901D (2022) https://doi.org/10.1117/12.2641143
Event: International Conference on Computer Network Security and Software Engineering (CNSSE 2022), 2022, Zhuhai, China

Abstract

In the semantic segmentation of images, the problems of incoherence of segmented objects, inconsistency of small objects, and mismatch between objects and labels have become the bottleneck restricting the application of semantic segmentation. Accordingly, an encoder-decoder network is designed with an improved residual network, and an improved model combining attention mechanism and pyramid module is proposed. Among them, the attention module enhances the position information of objects, and the pyramid module improves the segmentation accuracy of object outlines and small objects. The simulation experiments were carried out on the PASCAL VOC 2012 dataset, and the results showed that the MIoU (mean intersection over union, average intersection ratio) reached 79.5% in the improved network structure on the VOC 2012 dataset, which was higher than the accuracy of the basic algorithm. 2.8%, which verifies the effectiveness of the proposed method.

Citation Download Citation

Pengyu Zhang, Hongwei Li, Yashuai Zhao, and Pengfei Zhang "Improved semantic segmentation algorithm based on attention mechanism and pyramid module", Proc. SPIE 12290, International Conference on Computer Network Security and Software Engineering (CNSSE 2022), 122901D (3 October 2022); https://doi.org/10.1117/12.2641143

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