Title: Social media image classification and retrieval method based on deep hash algorithm

Authors: Zilong Li; Yong Zhou; Hongdong Wang

Addresses: School of Information Engineering, Xu'zhou University of Technology, Xu'zhou 221018, China; School of Computer Science and Technology, China University of Mining and Technology, Xu'zhou 221116, China; Post Doctoral Research Center, Onnes Power Technology Co., Ltd., Xu'zhou 221003, China ' School of Computer Science and Technology, China University of Mining and Technology, Xu'zhou 221116, China ' School of Information Engineering, Xu'zhou University of Technology, Xu'zhou 221018, China

Abstract: In order to solve the problems of large classification error and low retrieval accuracy of traditional retrieval methods, a social media image classification retrieval method based on deep hash algorithm is proposed. With the help of CBOW model to extract semantic features and colour histogram to extract colour features of social media images, social media image classification is completed. Hash algorithm is used to deal with the unidirectional irreversibility of social media image, and the pixel feature points in the image are regarded as density function for one-to-one correspondence. The loss function is used to control the convergence of the hash algorithm to achieve social media image retrieval. Experimental results show that the error of social media image classification is only 2%, the retrieval accuracy is always higher than 90%, and the retrieval time is only 3.4 s, which has the advantage of high retrieval efficiency.

Keywords: social media images; CBOW model; clustering algorithm; deep hash algorithm; loss function.

DOI: 10.1504/IJWBC.2022.125506

International Journal of Web Based Communities, 2022 Vol.18 No.3/4, pp.276 - 287

Received: 04 Jun 2021
Accepted: 05 Nov 2021
Published online: 12 Sep 2022 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article