A Novel Re-Ranking Based Image Retrieval Using Convolutional Neural Network and Support Vector Machine in Cloud Environment

Authors

  • K. Nithya, V. Rajamani

DOI:

https://doi.org/10.17762/msea.v71i3.205

Abstract

Content Based Image Retrieval (CBIR) is the prominent research area now-a-days. In spite of massive researches exist in the field of CBIR, the retrieval of relevant images from the large set of image database in the cloud, remains more challenging task. The real problem lies in the feature representation of images and semantic gap exists between the image representation as a pixel in the machine and the concepts viewed by the human. Among all the techniques, the Convolutional Neural Network (CNN) stands top in bridging this gap. In this paper, a deep CNN ResNet50 model is used to represent the features of the image. During training the CNN, the Support Vector Machine (SVM) is used in the place of Softmax layer in the CNN to predict the class label of the training image. To further improve the accuracy of the proposed model, a Novel re-ranking method based on K-Nearest Neighbourhood (KNN) is used. The proposed system exhibits better performance and accuracy than the existing systems.

Downloads

Published

2022-06-09

How to Cite

K. Nithya, V. Rajamani. (2022). A Novel Re-Ranking Based Image Retrieval Using Convolutional Neural Network and Support Vector Machine in Cloud Environment. Mathematical Statistician and Engineering Applications, 71(3), 665 –. https://doi.org/10.17762/msea.v71i3.205

Issue

Section

Articles