A convolutional neural network based complex scene classification framework using transfer deep combined convolutional activations

Shuyun Liu*, Hong Wang, Yutong Jiang, Zhonglin Yang, Zhiyang Ma

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

In many scene classification applications, the variety of surface objects, high within-category diversity and between-category similarity carry challenges for the classification Framework. Most of CNN-based classification methods only extract image features from a single network layer, which may cause the completed image information difficult to extract in complex scenes. We propose a novel transfer deep combined convolutional activations (TDCCA) to integrate both the low-level and high-level features. Extensive comparative experiments are conducted on UC Merced database, Aerial Image database and NWPU-RESISC45 database. The results reveal that our proposed TDCCA achieves higher experimental accuracies than other up-to-date popular methods.

Original languageEnglish
Title of host publication2021 International Conference on Optical Instruments and Technology
Subtitle of host publicationOptical Systems, Optoelectronic Instruments, Novel Display, and Imaging Technology
EditorsJuan Liu, Baohua Jia, Liangcai Cao, Xincheng Yao, Yongtian Wang, Takanori Nomura
PublisherSPIE
ISBN (Electronic)9781510655591
DOIs
Publication statusPublished - 2022
Event2021 International Conference on Optical Instruments and Technology: Optical Systems, Optoelectronic Instruments, Novel Display, and Imaging Technology - Virtual, Online, China
Duration: 8 Apr 202210 Apr 2022

Publication series

NameProceedings of SPIE - The International Society for Optical Engineering
Volume12277
ISSN (Print)0277-786X
ISSN (Electronic)1996-756X

Conference

Conference2021 International Conference on Optical Instruments and Technology: Optical Systems, Optoelectronic Instruments, Novel Display, and Imaging Technology
Country/TerritoryChina
CityVirtual, Online
Period8/04/2210/04/22

Keywords

  • Transfer deep combined convolutional activations
  • convolutional neural network
  • image scene classification
  • weighted k-nearest neighbor classifier

Fingerprint

Dive into the research topics of 'A convolutional neural network based complex scene classification framework using transfer deep combined convolutional activations'. Together they form a unique fingerprint.

Cite this