Cargando…

Multipath Lightweight Deep Network Using Randomly Selected Dilated Convolution

Robot vision is an essential research field that enables machines to perform various tasks by classifying/detecting/segmenting objects as humans do. The classification accuracy of machine learning algorithms already exceeds that of a well-trained human, and the results are rather saturated. Hence, i...

Descripción completa

Detalles Bibliográficos
Autores principales: Park, Sangun, Chang, Dong Eui
Formato: Online Artículo Texto
Lenguaje:English
Publicado: MDPI 2021
Materias:
Acceso en línea:https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8659751/
https://www.ncbi.nlm.nih.gov/pubmed/34883865
http://dx.doi.org/10.3390/s21237862
_version_ 1784613038898282496
author Park, Sangun
Chang, Dong Eui
author_facet Park, Sangun
Chang, Dong Eui
author_sort Park, Sangun
collection PubMed
description Robot vision is an essential research field that enables machines to perform various tasks by classifying/detecting/segmenting objects as humans do. The classification accuracy of machine learning algorithms already exceeds that of a well-trained human, and the results are rather saturated. Hence, in recent years, many studies have been conducted in the direction of reducing the weight of the model and applying it to mobile devices. For this purpose, we propose a multipath lightweight deep network using randomly selected dilated convolutions. The proposed network consists of two sets of multipath networks (minimum 2, maximum 8), where the output feature maps of one path are concatenated with the input feature maps of the other path so that the features are reusable and abundant. We also replace the [Formula: see text] standard convolution of each path with a randomly selected dilated convolution, which has the effect of increasing the receptive field. The proposed network lowers the number of floating point operations (FLOPs) and parameters by more than 50% and the classification error by 0.8% as compared to the state-of-the-art. We show that the proposed network is efficient.
format Online
Article
Text
id pubmed-8659751
institution National Center for Biotechnology Information
language English
publishDate 2021
publisher MDPI
record_format MEDLINE/PubMed
spelling pubmed-86597512021-12-10 Multipath Lightweight Deep Network Using Randomly Selected Dilated Convolution Park, Sangun Chang, Dong Eui Sensors (Basel) Article Robot vision is an essential research field that enables machines to perform various tasks by classifying/detecting/segmenting objects as humans do. The classification accuracy of machine learning algorithms already exceeds that of a well-trained human, and the results are rather saturated. Hence, in recent years, many studies have been conducted in the direction of reducing the weight of the model and applying it to mobile devices. For this purpose, we propose a multipath lightweight deep network using randomly selected dilated convolutions. The proposed network consists of two sets of multipath networks (minimum 2, maximum 8), where the output feature maps of one path are concatenated with the input feature maps of the other path so that the features are reusable and abundant. We also replace the [Formula: see text] standard convolution of each path with a randomly selected dilated convolution, which has the effect of increasing the receptive field. The proposed network lowers the number of floating point operations (FLOPs) and parameters by more than 50% and the classification error by 0.8% as compared to the state-of-the-art. We show that the proposed network is efficient. MDPI 2021-11-26 /pmc/articles/PMC8659751/ /pubmed/34883865 http://dx.doi.org/10.3390/s21237862 Text en © 2021 by the authors. https://creativecommons.org/licenses/by/4.0/Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
spellingShingle Article
Park, Sangun
Chang, Dong Eui
Multipath Lightweight Deep Network Using Randomly Selected Dilated Convolution
title Multipath Lightweight Deep Network Using Randomly Selected Dilated Convolution
title_full Multipath Lightweight Deep Network Using Randomly Selected Dilated Convolution
title_fullStr Multipath Lightweight Deep Network Using Randomly Selected Dilated Convolution
title_full_unstemmed Multipath Lightweight Deep Network Using Randomly Selected Dilated Convolution
title_short Multipath Lightweight Deep Network Using Randomly Selected Dilated Convolution
title_sort multipath lightweight deep network using randomly selected dilated convolution
topic Article
url https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8659751/
https://www.ncbi.nlm.nih.gov/pubmed/34883865
http://dx.doi.org/10.3390/s21237862
work_keys_str_mv AT parksangun multipathlightweightdeepnetworkusingrandomlyselecteddilatedconvolution
AT changdongeui multipathlightweightdeepnetworkusingrandomlyselecteddilatedconvolution