Performance evaluation on quantized weight for convolutional neural network based object detection

A Convolutional Neural Network (CNN) based object detection is an emerging topic in the image processing field and has become the state-of-the-art in computer vision and machine learning. The traditional system in object detection uses a handcrafted feature extractor which is less robust in applicat...

Full description

Saved in:

Bibliographic Details
Main Author:	Putra, Mohd Hasbullah
Format:	Thesis
Language:	English English
Published:	2021
Subjects:	T Technology (General)
Online Access:	http://eprints.utem.edu.my/id/eprint/26088/1/Performance%20evaluation%20on%20quantized%20weight%20for%20convolutional%20neural%20network%20based%20object%20detection.pdf http://eprints.utem.edu.my/id/eprint/26088/2/Performance%20evaluation%20on%20quantized%20weight%20for%20convolutional%20neural%20network%20based%20object%20detection.pdf
Tags:	Add Tag No Tags, Be the first to tag this record!

id	my-utem-ep.26088
record_format	uketd_dc
spelling	my-utem-ep.260882023-01-13T15:50:13Z Performance evaluation on quantized weight for convolutional neural network based object detection 2021 Putra, Mohd Hasbullah T Technology (General) TK Electrical engineering. Electronics Nuclear engineering A Convolutional Neural Network (CNN) based object detection is an emerging topic in the image processing field and has become the state-of-the-art in computer vision and machine learning. The traditional system in object detection uses a handcrafted feature extractor which is less robust in applications. By applying the CNN approach in the field, the accuracy of object detections can increase significantly. However, the use of deep CNN architecture model leads to high computation. In this research, a real-time CNN based object detection system is presented. The system is designed based on the modified You Only Look Once (modified-YOLO) architecture which is constructed with only 7 CNN layers. The grid cell parameter value of the system is varied to evaluate its effectiveness and ability in detecting small size objects upon deployment. The experimental results demonstrate that even with 7 convolutional layers, modified-YOLO can provide good detection accuracy and real-time operation achieving the best miss rate (MR) of 22.7% MR. Although the scores show an increase in the MR, the visual qualitative evaluation using randomly captured images indicate that the 7 layers modified-YOLO architecture with 11x11 grid cells can correctly and easily detect small objects. This makes the modified YOLO architecture which has been reduced in terms of complexity a suitable candidate for use in real-time operation. In order to further reduce the complexity of the CNN system, the trained floating-point weights are quantized. Three types of scalar quantization are used to quantize the CNN weights namely symmetric uniform quantizer, asymmetric uniform quantizer and non-uniform quantizer designed using k-means algorithm. The quantization reduces the storage and computation requirements. The quantitative results using the MR standard metric indicate that the non-uniform quantizer provides the best results compared to the other quantization methods. Using 6-bit precision non-uniformly quantized weights yields detection performance comparable to the CNN network using floating-point weights. Additionally, based on the qualitative results, the CNN network with 4-bit non-uniform quantization weights is able to detect the person objects correctly. 2021 Thesis http://eprints.utem.edu.my/id/eprint/26088/ http://eprints.utem.edu.my/id/eprint/26088/1/Performance%20evaluation%20on%20quantized%20weight%20for%20convolutional%20neural%20network%20based%20object%20detection.pdf text en public http://eprints.utem.edu.my/id/eprint/26088/2/Performance%20evaluation%20on%20quantized%20weight%20for%20convolutional%20neural%20network%20based%20object%20detection.pdf text en validuser https://plh.utem.edu.my/cgi-bin/koha/opac-detail.pl?biblionumber=121303 dphil masters Universiti Teknikal Malaysia Melaka Faculty of Electronic and Computer Engineering Mohd Yussof, Zulkalnain
institution	Universiti Teknikal Malaysia Melaka
collection	UTeM Repository
language	English English
advisor	Mohd Yussof, Zulkalnain
topic	T Technology (General) T Technology (General)
spellingShingle	T Technology (General) T Technology (General) Putra, Mohd Hasbullah Performance evaluation on quantized weight for convolutional neural network based object detection
description	A Convolutional Neural Network (CNN) based object detection is an emerging topic in the image processing field and has become the state-of-the-art in computer vision and machine learning. The traditional system in object detection uses a handcrafted feature extractor which is less robust in applications. By applying the CNN approach in the field, the accuracy of object detections can increase significantly. However, the use of deep CNN architecture model leads to high computation. In this research, a real-time CNN based object detection system is presented. The system is designed based on the modified You Only Look Once (modified-YOLO) architecture which is constructed with only 7 CNN layers. The grid cell parameter value of the system is varied to evaluate its effectiveness and ability in detecting small size objects upon deployment. The experimental results demonstrate that even with 7 convolutional layers, modified-YOLO can provide good detection accuracy and real-time operation achieving the best miss rate (MR) of 22.7% MR. Although the scores show an increase in the MR, the visual qualitative evaluation using randomly captured images indicate that the 7 layers modified-YOLO architecture with 11x11 grid cells can correctly and easily detect small objects. This makes the modified YOLO architecture which has been reduced in terms of complexity a suitable candidate for use in real-time operation. In order to further reduce the complexity of the CNN system, the trained floating-point weights are quantized. Three types of scalar quantization are used to quantize the CNN weights namely symmetric uniform quantizer, asymmetric uniform quantizer and non-uniform quantizer designed using k-means algorithm. The quantization reduces the storage and computation requirements. The quantitative results using the MR standard metric indicate that the non-uniform quantizer provides the best results compared to the other quantization methods. Using 6-bit precision non-uniformly quantized weights yields detection performance comparable to the CNN network using floating-point weights. Additionally, based on the qualitative results, the CNN network with 4-bit non-uniform quantization weights is able to detect the person objects correctly.
format	Thesis
qualification_name	Doctor of Philosophy (PhD.)
qualification_level	Master's degree
author	Putra, Mohd Hasbullah
author_facet	Putra, Mohd Hasbullah
author_sort	Putra, Mohd Hasbullah
title	Performance evaluation on quantized weight for convolutional neural network based object detection
title_short	Performance evaluation on quantized weight for convolutional neural network based object detection
title_full	Performance evaluation on quantized weight for convolutional neural network based object detection
title_fullStr	Performance evaluation on quantized weight for convolutional neural network based object detection
title_full_unstemmed	Performance evaluation on quantized weight for convolutional neural network based object detection
title_sort	performance evaluation on quantized weight for convolutional neural network based object detection
granting_institution	Universiti Teknikal Malaysia Melaka
granting_department	Faculty of Electronic and Computer Engineering
publishDate	2021
url	http://eprints.utem.edu.my/id/eprint/26088/1/Performance%20evaluation%20on%20quantized%20weight%20for%20convolutional%20neural%20network%20based%20object%20detection.pdf http://eprints.utem.edu.my/id/eprint/26088/2/Performance%20evaluation%20on%20quantized%20weight%20for%20convolutional%20neural%20network%20based%20object%20detection.pdf
_version_	1776103135572393984

Performance evaluation on quantized weight for convolutional neural network based object detection

Similar Items