Publication detail

Image Background Noise Impact on Convolutional Neural Network Training

RAJNOHA, M. BURGET, R. POVODA, L.

Original Title

Image Background Noise Impact on Convolutional Neural Network Training

Type

conference paper

Language

English

Original Abstract

Small size dataset is general issue that we may encounter when training neural networks for analysis of image data. There are many cases when networks can not start training even with data augmentation. This paper proposes a new method how to allow training of image classification even when traditional approaches fail. It presents an experiment, which shows that subtraction of redundant background from images can significantly improve convergence of neural network training. Improvement is not in accuracy matter but it means that neural network is able to train and to start convergence. For experimental evaluation, person binary classification was used and compared to experiments, where the background was removed.

Keywords

snr; signal to noise ratio; background; noise; training; impact; cnn; convolutional neural networks; small dataset

Authors

RAJNOHA, M.; BURGET, R.; POVODA, L.

Released

6. 11. 2018

Location

Moskva

ISBN

978-1-5386-9361-2

Book

2018 10th International Congress on Ultra Modern Telecommunications and Control Systems and Workshops (ICUMT)

Pages from

168

Pages to

171

Pages count

4

URL

https://ieeexplore.ieee.org/document/8631242

BibTex

@inproceedings{BUT150877,
  author="Martin {Rajnoha} and Radim {Burget} and Lukáš {Povoda}",
  title="Image Background Noise Impact on Convolutional Neural Network Training",
  booktitle="2018 10th International Congress on Ultra Modern Telecommunications and Control Systems
and Workshops (ICUMT)",
  year="2018",
  pages="168--171",
  address="Moskva",
  doi="10.1109/ICUMT.2018.8631242",
  isbn="978-1-5386-9361-2",
  url="https://ieeexplore.ieee.org/document/8631242"
}