Cover Image

An improved object detection model based on optimised CNN.

Object detection is a computer vision technique that gives the ability to individually locate, recognise, and interpret multiple objects in an image with a better understanding. Modern image understanding tasks like image classification have been improved by state-of-the-art deep learning methods, p...

Full description

Saved in:
Bibliographic Details
Main Authors: Mohd. Anuar, Mohd. Syahid, Jayapalan, Senthil Kumar
Format: Article
Language:English
Published: Penerbit UTM Press 2022
Subjects:
TJ Mechanical engineering and machinery
Online Access:http://eprints.utm.my/104593/1/SenthilKumarJayapalanSyahidAnuar2022_AnImprovedObjectDetectionModel.pdf
http://eprints.utm.my/104593/
https://oiji.utm.my/index.php/oiji/article/view/230
Tags: Add Tag
No Tags, Be the first to tag this record!
id my.utm.104593
record_format eprints
spelling my.utm.1045932024-02-21T08:21:59Z http://eprints.utm.my/104593/ An improved object detection model based on optimised CNN. Mohd. Anuar, Mohd. Syahid Jayapalan, Senthil Kumar TJ Mechanical engineering and machinery Object detection is a computer vision technique that gives the ability to individually locate, recognise, and interpret multiple objects in an image with a better understanding. Modern image understanding tasks like image classification have been improved by state-of-the-art deep learning methods, particularly by convolutional neural networks (CNN). Region-based object detection algorithms such as Fast-RCNN achieve classification by CNN but over a longer period of time. You only look once (YOLO) prompts the object location and classification, treating object detection as a regression problem in an end-to-end network in a single step, whereas its accuracy decreases when the image has similar objects in a confined area, particularly when independent of the surrounding context. The aim of the current study is to improve YOLOv3 by optimising Darknet-53 to address the memory issue, using switchable normalisation techniques. We investigated the performance of five pre-trained networks, SqueezeNet, GoogleNet, ShuffleNet, Darknet-53, and Inception-V3, using a confusion matrix employing various epochs, learning rates, and mini-batches based on transfer learning. Darknet-53 took five times longer to complete the training and also ran into errors, most likely due to GPU memory shortages, whereas GoogleNet virtually obtained the same results in a fraction of the time. Using switchable normalisation techniques with the 10 class CIFAR-10 dataset, and utilising deep network designer (DND) of MATLAB R2021a, optimised versions of Darknet-53 increased the validation accuracy, considerably reducing the training time, and rectified the memory issue, which were then used as a backbone for YOLOv3 for effective object detection. The enhanced YOLOv3 was then assessed using a vehicle dataset and a sample Kuala Lumpur traffic scene using average precision. YOLOv3 with optimised CNN dNet-CIN as the backbone produced the best experimental results, with an FPS of 3.21 and a mAP-50 of 97%. Penerbit UTM Press 2022-12-15 Article PeerReviewed application/pdf en http://eprints.utm.my/104593/1/SenthilKumarJayapalanSyahidAnuar2022_AnImprovedObjectDetectionModel.pdf Mohd. Anuar, Mohd. Syahid and Jayapalan, Senthil Kumar (2022) An improved object detection model based on optimised CNN. Open International Journal Of Informatics, 10 (1). pp. 78-96. ISSN 2289-2370 https://oiji.utm.my/index.php/oiji/article/view/230 NA
institution Universiti Teknologi Malaysia
building UTM Library
collection Institutional Repository
continent Asia
country Malaysia
content_provider Universiti Teknologi Malaysia
content_source UTM Institutional Repository
url_provider http://eprints.utm.my/
language English
topic TJ Mechanical engineering and machinery
spellingShingle TJ Mechanical engineering and machinery
Mohd. Anuar, Mohd. Syahid
Jayapalan, Senthil Kumar
An improved object detection model based on optimised CNN.
description Object detection is a computer vision technique that gives the ability to individually locate, recognise, and interpret multiple objects in an image with a better understanding. Modern image understanding tasks like image classification have been improved by state-of-the-art deep learning methods, particularly by convolutional neural networks (CNN). Region-based object detection algorithms such as Fast-RCNN achieve classification by CNN but over a longer period of time. You only look once (YOLO) prompts the object location and classification, treating object detection as a regression problem in an end-to-end network in a single step, whereas its accuracy decreases when the image has similar objects in a confined area, particularly when independent of the surrounding context. The aim of the current study is to improve YOLOv3 by optimising Darknet-53 to address the memory issue, using switchable normalisation techniques. We investigated the performance of five pre-trained networks, SqueezeNet, GoogleNet, ShuffleNet, Darknet-53, and Inception-V3, using a confusion matrix employing various epochs, learning rates, and mini-batches based on transfer learning. Darknet-53 took five times longer to complete the training and also ran into errors, most likely due to GPU memory shortages, whereas GoogleNet virtually obtained the same results in a fraction of the time. Using switchable normalisation techniques with the 10 class CIFAR-10 dataset, and utilising deep network designer (DND) of MATLAB R2021a, optimised versions of Darknet-53 increased the validation accuracy, considerably reducing the training time, and rectified the memory issue, which were then used as a backbone for YOLOv3 for effective object detection. The enhanced YOLOv3 was then assessed using a vehicle dataset and a sample Kuala Lumpur traffic scene using average precision. YOLOv3 with optimised CNN dNet-CIN as the backbone produced the best experimental results, with an FPS of 3.21 and a mAP-50 of 97%.
format Article
author Mohd. Anuar, Mohd. Syahid
Jayapalan, Senthil Kumar
author_facet Mohd. Anuar, Mohd. Syahid
Jayapalan, Senthil Kumar
author_sort Mohd. Anuar, Mohd. Syahid
title An improved object detection model based on optimised CNN.
title_short An improved object detection model based on optimised CNN.
title_full An improved object detection model based on optimised CNN.
title_fullStr An improved object detection model based on optimised CNN.
title_full_unstemmed An improved object detection model based on optimised CNN.
title_sort improved object detection model based on optimised cnn.
publisher Penerbit UTM Press
publishDate 2022
url http://eprints.utm.my/104593/1/SenthilKumarJayapalanSyahidAnuar2022_AnImprovedObjectDetectionModel.pdf
http://eprints.utm.my/104593/
https://oiji.utm.my/index.php/oiji/article/view/230
_version_ 1792147850075832320
score 13.211869

Similar Items