IRMA-International.org: Creator of Knowledge
Information Resources Management Association
Advancing the Concepts & Practices of Information Resources Management in Modern Organizations

Analysis of Image Similarity Using CNN and ANNOY

Analysis of Image Similarity Using CNN and ANNOY
View Sample PDF
Author(s): Jun-Ki Hong (Pai Chai University, South Korea)
Copyright: 2022
Volume: 10
Issue: 2
Pages: 11
Source title: International Journal of Software Innovation (IJSI)
Editor(s)-in-Chief: Roger Y. Lee (Central Michigan University, USA) and Lawrence Chung (The University of Texas at Dallas, USA)
DOI: 10.4018/IJSI.289593

Purchase

View Analysis of Image Similarity Using CNN and ANNOY on the publisher's website for pricing and purchasing information.

Abstract

This article proposes an algorithm to more efficiently search for clothing product images that are similar to a new input clothing product image. Convolutional Neural Network (CNN) and Artificial Neural Network Oh Yeah (ANNOY) technologies were applied to a database of 60,000 clothing images, and the similarity and the processing rates of the two technologies were compared. The conventional CNN technology searches similar images by exploring all the pixels of an image, while the ANNOY technology uses a binary tree node, which is the similarity distance measured between images. The ANNOY technology can drastically reduce image search time, although the image similarity accuracy is slightly decreased. The reduction in image search time saves costs, and the rapid search processing rate enables the technology to be applied to various kinds of online services, including product search, product comparison and product recommendation.

Related Content

SunMyung Hwang, Hee Gyun Yeom. © 2022. 10 pages.
Jun-Ki Hong. © 2022. 11 pages.
Ha Jin Hwang, Haeng Kon Kim, Monowar Mahmood, Norazryana Mat Dawi. © 2022. 11 pages.
Sang-Kwon Yun, Hye Jeong Kwon, Jongbae Kim. © 2022. 13 pages.
Sung Hwa Han, Min Hye Jwa, Sang Bin Jeong, Gwangyong Gim. © 2022. 13 pages.
Euntack Im, Dukjin Kim, Minhye Jwa, Gwangyong Gim. © 2022. 13 pages.
SungKwang Kim, YoungHwan Im. © 2022. 11 pages.
Body Bottom