Discussion on the degree of influence on the performance of scene semantic classification after applying SRCNN reconstruction model

Liu Kexin; Liu Siyuan

doi:10.1117/12.2628423

21 April 2022 Discussion on the degree of influence on the performance of scene semantic classification after applying SRCNN reconstruction model

Liu Kexin, Liu Siyuan

Author Affiliations +

Proceedings Volume 12175, International Conference on Network Communication and Information Security (ICNCIS 2021); 121750H (2022) https://doi.org/10.1117/12.2628423
Event: International Conference on Network Communication and Information Security (ICNCIS 2021), 2021, Beijing, China

Abstract

Scene images are rich in semantics, so scene classification is a very valuable and challenging task. Deep learning methods represented by DCNN have quickly become research hotspots with high efficiency and high accuracy. In response to the problem of improving classification performance, a network model that optimizes initial weights has appeared since 2018. However, none of the existing methods takes into account the adverse effect of the low quality of network input images on the classification accuracy. This article will combine super-resolution reconstruction technology to add image preprocessing operations before the classification network. The specific performance is the introduction of the SRCNN reconstruction model on the basis of the DCNN-ELM classification network. The following four experiments are implemented for three types of data sets: 1. Taking MIT Indoor and LSP as the research objects, compare the classification performance of the six existing methods to verify the multi-applicability of the optimized classification model. 2. Take five types of news scene sets as the research object to test the robustness of the optimized classification model. 3. Select three types of news scenes as the research objects, and divide them according to the degree of semantic clarity. The feasibility of the classification method proposed in this paper is discussed through the classification comparison experiment of whether the reconstruction preprocessing is performed on the input image. 4. Taking the atlas collected under severe weather as the research object, the application potential of the method in this paper is illustrated through the abovementioned comparative experiments.

Citation Download Citation

Liu Kexin and Liu Siyuan "Discussion on the degree of influence on the performance of scene semantic classification after applying SRCNN reconstruction model", Proc. SPIE 12175, International Conference on Network Communication and Information Security (ICNCIS 2021), 121750H (21 April 2022); https://doi.org/10.1117/12.2628423

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

Members: $17.00

Non-members: $21.00 ADD TO CART

PROCEEDINGS
9 PAGES

DOWNLOAD PAPER SAVE TO MY LIBRARY

GET CITATION

RIGHTS & PERMISSIONS

Get copyright permission Get copyright permission on Copyright Marketplace

KEYWORDS

Classification systems

Scene classification

Image classification

Floods

Performance modeling

Data modeling

Visualization

Show All Keywords

Keywords/Phrases

Search In:

Publication Years