Abstract:

The growing increase of demand in computer vision, arises the problem of handling specific algorithmic to this field. Image classification is one of these problems widely used in many applications. Starting from late 2000’s Neural Networks based algorithms perform best in this task. The most accurate results are obtained with help of Convolutional Neural Networks, which currently outperform all other algorithms, at least by accuracy. Nevertheless, they have problems connected with the memory use and training time, which in some cases are not acceptable. Thus there is a need for techniques for tackling with problem and fining tradeoff between complexity, time and accuracy.

Publisher:

ITHEA

Date of publication:

2019

Identifier:

oai:noad.sci.am:135926

ISSN:

1314-6416

Language:

English

Journal or Publication Title:

Information Models and Analyses

Volume:

8

Number:

1

URL:

Additional Information:

mkrthcyan.karlen@icloud.com

Affiliation:

Institute for Informatics and Automation Problems

Country:

Armenia