Evaluation of Deep Neural Network Compression Methods for Edge Devices Using Weighted Score-Based Ranking Scheme

The demand for object detection capability in edge computing systems has surged. As such, the need for lightweight Convolutional Neural Network (CNN)-based object detection models has become a focal point. Current models are large in memory and deployment in edge devices is demanding. This shows tha...

Descripción completa

Guardado en:
Detalles Bibliográficos
Autores principales: Olutosin Ajibola Ademola, Mairo Leier, Eduard Petlenkov
Formato: article
Lenguaje:EN
Publicado: MDPI AG 2021
Materias:
Acceso en línea:https://doaj.org/article/348a22621e7447a08b4c016aa14abe64
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
id oai:doaj.org-article:348a22621e7447a08b4c016aa14abe64
record_format dspace
spelling oai:doaj.org-article:348a22621e7447a08b4c016aa14abe642021-11-25T18:57:15ZEvaluation of Deep Neural Network Compression Methods for Edge Devices Using Weighted Score-Based Ranking Scheme10.3390/s212275291424-8220https://doaj.org/article/348a22621e7447a08b4c016aa14abe642021-11-01T00:00:00Zhttps://www.mdpi.com/1424-8220/21/22/7529https://doaj.org/toc/1424-8220The demand for object detection capability in edge computing systems has surged. As such, the need for lightweight Convolutional Neural Network (CNN)-based object detection models has become a focal point. Current models are large in memory and deployment in edge devices is demanding. This shows that the models need to be optimized for the hardware without performance degradation. There exist several model compression methods; however, determining the most efficient method is of major concern. Our goal was to rank the performance of these methods using our application as a case study. We aimed to develop a real-time vehicle tracking system for cargo ships. To address this, we developed a weighted score-based ranking scheme that utilizes the model performance metrics. We demonstrated the effectiveness of this method by applying it on the baseline, compressed, and micro-CNN models trained on our dataset. The result showed that quantization is the most efficient compression method for the application, having the highest rank, with an average weighted score of 9.00, followed by binarization, having an average weighted score of 8.07. Our proposed method is extendable and can be used as a framework for the selection of suitable model compression methods for edge devices in different applications.Olutosin Ajibola AdemolaMairo LeierEduard PetlenkovMDPI AGarticledeep neural network compressioncompression method evaluationweighted score-based rankingembedded deep learningedge computingChemical technologyTP1-1185ENSensors, Vol 21, Iss 7529, p 7529 (2021)
institution DOAJ
collection DOAJ
language EN
topic deep neural network compression
compression method evaluation
weighted score-based ranking
embedded deep learning
edge computing
Chemical technology
TP1-1185
spellingShingle deep neural network compression
compression method evaluation
weighted score-based ranking
embedded deep learning
edge computing
Chemical technology
TP1-1185
Olutosin Ajibola Ademola
Mairo Leier
Eduard Petlenkov
Evaluation of Deep Neural Network Compression Methods for Edge Devices Using Weighted Score-Based Ranking Scheme
description The demand for object detection capability in edge computing systems has surged. As such, the need for lightweight Convolutional Neural Network (CNN)-based object detection models has become a focal point. Current models are large in memory and deployment in edge devices is demanding. This shows that the models need to be optimized for the hardware without performance degradation. There exist several model compression methods; however, determining the most efficient method is of major concern. Our goal was to rank the performance of these methods using our application as a case study. We aimed to develop a real-time vehicle tracking system for cargo ships. To address this, we developed a weighted score-based ranking scheme that utilizes the model performance metrics. We demonstrated the effectiveness of this method by applying it on the baseline, compressed, and micro-CNN models trained on our dataset. The result showed that quantization is the most efficient compression method for the application, having the highest rank, with an average weighted score of 9.00, followed by binarization, having an average weighted score of 8.07. Our proposed method is extendable and can be used as a framework for the selection of suitable model compression methods for edge devices in different applications.
format article
author Olutosin Ajibola Ademola
Mairo Leier
Eduard Petlenkov
author_facet Olutosin Ajibola Ademola
Mairo Leier
Eduard Petlenkov
author_sort Olutosin Ajibola Ademola
title Evaluation of Deep Neural Network Compression Methods for Edge Devices Using Weighted Score-Based Ranking Scheme
title_short Evaluation of Deep Neural Network Compression Methods for Edge Devices Using Weighted Score-Based Ranking Scheme
title_full Evaluation of Deep Neural Network Compression Methods for Edge Devices Using Weighted Score-Based Ranking Scheme
title_fullStr Evaluation of Deep Neural Network Compression Methods for Edge Devices Using Weighted Score-Based Ranking Scheme
title_full_unstemmed Evaluation of Deep Neural Network Compression Methods for Edge Devices Using Weighted Score-Based Ranking Scheme
title_sort evaluation of deep neural network compression methods for edge devices using weighted score-based ranking scheme
publisher MDPI AG
publishDate 2021
url https://doaj.org/article/348a22621e7447a08b4c016aa14abe64
work_keys_str_mv AT olutosinajibolaademola evaluationofdeepneuralnetworkcompressionmethodsforedgedevicesusingweightedscorebasedrankingscheme
AT mairoleier evaluationofdeepneuralnetworkcompressionmethodsforedgedevicesusingweightedscorebasedrankingscheme
AT eduardpetlenkov evaluationofdeepneuralnetworkcompressionmethodsforedgedevicesusingweightedscorebasedrankingscheme
_version_ 1718410515292618752