当前位置: X-MOL 学术Concurr. Comput. Pract. Exp. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
A new method for image classification and image retrieval using convolutional neural networks
Concurrency and Computation: Practice and Experience ( IF 1.5 ) Pub Date : 2021-07-28 , DOI: 10.1002/cpe.6533
Davar Giveki 1 , Ashkan Shakarami 2 , Hadis Tarrah 3 , Mohammad Ali Soltanshahi 4
Affiliation  

This article proposes a new method for image classification and image retrieval. The advantages of the proposed method are its high performance and requiring less memory compared to other methods. In order to extract image features, a Convolutional Neural Network (CNN), AlexNet, has been used. For image classification, we design a committee of four classifiers trained on graphics cards, narrowing the gap to human performance. For image retrieval, the similarity between extracted features from dataset images and features of the query image is calculated and the final results are visualized. Comprehensive experiments on Corel-1k, Corel-10k, Caltech-101 object and Scene-67 datasets have been investigated to find optimal parameters of the proposed method. The experiments demonstrate the high performance of the proposed method in comparison with the state-of-the-art in the field.

中文翻译:

一种使用卷积神经网络进行图像分类和图像检索的新方法

本文提出了一种新的图像分类和图像检索方法。与其他方法相比,所提出的方法的优点是其高性能和需要更少的内存。为了提取图像特征,使用了卷积神经网络 (CNN) AlexNet。对于图像分类,我们设计了一个由四个在显卡上训练的分类器组成的委员会,缩小了与人类性能的差距。对于图像检索,计算从数据集图像中提取的特征与查询图像的特征之间的相似度,并将最终结果可视化。已经研究了 Corel-1k、Corel-10k、Caltech-101 对象和 Scene-67 数据集的综合实验,以找到所提出方法的最佳参数。
更新日期:2021-07-28
down
wechat
bug