Detection method of potato leaf disease based on YOLOv5s

An improved leaf target detection method based on the YOLOv5s network is proposed to address the issues of low model detection accuracy and slow detection speed in potato leaf image target detection. Firstly, a deformable convolution replaces the standard convolution in YOLOv5s to ensure that the c...

Full description

Saved in:
Bibliographic Details
Main Authors: Jingtao Li, Hao Chen, Guisong Li, Yueqi Liu, Yanli Yang, Xia Liu, Chang Yi Wang
Format: Article
Language:English
Published: PAGEPress Publications 2024-06-01
Series:Journal of Agricultural Engineering
Subjects:
Online Access:https://www.agroengineering.org/jae/article/view/1587
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850149213432709120
author Jingtao Li
Hao Chen
Guisong Li
Yueqi Liu
Yanli Yang
Xia Liu
Chang Yi Wang
author_facet Jingtao Li
Hao Chen
Guisong Li
Yueqi Liu
Yanli Yang
Xia Liu
Chang Yi Wang
author_sort Jingtao Li
collection DOAJ
description An improved leaf target detection method based on the YOLOv5s network is proposed to address the issues of low model detection accuracy and slow detection speed in potato leaf image target detection. Firstly, a deformable convolution replaces the standard convolution in YOLOv5s to ensure that the convolution region always covers the target region. Secondly, CBAM attention module is introduced into the convolutional module to enhance local feature extraction and fusion capability of the network, while WIoU_Loss serves as Bounding box loss function SRN-DeblurNet deblurnet is combined with YOLOv5s network to convert part of fuzzy images into clear ones before being integrated with multi-scale features for model prediction. To verify its effectiveness, we trained our model using Pytorch deep learning framework and achieved an accuracy rate of 90.3% and recall rate of 88%, which are respectively 8.6% and 8.9% higher than those obtained by YOLOv5s.
format Article
id doaj-art-8967a0f185df41e39ccda8f36dc73ae4
institution OA Journals
issn 1974-7071
2239-6268
language English
publishDate 2024-06-01
publisher PAGEPress Publications
record_format Article
series Journal of Agricultural Engineering
spelling doaj-art-8967a0f185df41e39ccda8f36dc73ae42025-08-20T02:26:59ZengPAGEPress PublicationsJournal of Agricultural Engineering1974-70712239-62682024-06-01553Detection method of potato leaf disease based on YOLOv5sJingtao Li0Hao Chen1Guisong Li2Yueqi Liu3Yanli Yang4Xia Liu5Chang Yi Wang6School of Information Engineering and Automation, Kunming University of Science and TechnologySchool of Information Engineering and Automation, Kunming University of Science and TechnologySchool of Information Engineering and Automation, Kunming University of Science and TechnologySchool of Information Engineering and Automation, Kunming University of Science and TechnologyCollege of Plant Protection, Yunnan Agricultural University, KunmingCollege of Plant Protection, Yunnan Agricultural University, KunmingSchool of Animation and Digital Arts, Communication University of China, Beijing An improved leaf target detection method based on the YOLOv5s network is proposed to address the issues of low model detection accuracy and slow detection speed in potato leaf image target detection. Firstly, a deformable convolution replaces the standard convolution in YOLOv5s to ensure that the convolution region always covers the target region. Secondly, CBAM attention module is introduced into the convolutional module to enhance local feature extraction and fusion capability of the network, while WIoU_Loss serves as Bounding box loss function SRN-DeblurNet deblurnet is combined with YOLOv5s network to convert part of fuzzy images into clear ones before being integrated with multi-scale features for model prediction. To verify its effectiveness, we trained our model using Pytorch deep learning framework and achieved an accuracy rate of 90.3% and recall rate of 88%, which are respectively 8.6% and 8.9% higher than those obtained by YOLOv5s. https://www.agroengineering.org/jae/article/view/1587Potato leaf diseasetargeted detectionYOLOv5sdeformable convolution
spellingShingle Jingtao Li
Hao Chen
Guisong Li
Yueqi Liu
Yanli Yang
Xia Liu
Chang Yi Wang
Detection method of potato leaf disease based on YOLOv5s
Journal of Agricultural Engineering
Potato leaf disease
targeted detection
YOLOv5s
deformable convolution
title Detection method of potato leaf disease based on YOLOv5s
title_full Detection method of potato leaf disease based on YOLOv5s
title_fullStr Detection method of potato leaf disease based on YOLOv5s
title_full_unstemmed Detection method of potato leaf disease based on YOLOv5s
title_short Detection method of potato leaf disease based on YOLOv5s
title_sort detection method of potato leaf disease based on yolov5s
topic Potato leaf disease
targeted detection
YOLOv5s
deformable convolution
url https://www.agroengineering.org/jae/article/view/1587
work_keys_str_mv AT jingtaoli detectionmethodofpotatoleafdiseasebasedonyolov5s
AT haochen detectionmethodofpotatoleafdiseasebasedonyolov5s
AT guisongli detectionmethodofpotatoleafdiseasebasedonyolov5s
AT yueqiliu detectionmethodofpotatoleafdiseasebasedonyolov5s
AT yanliyang detectionmethodofpotatoleafdiseasebasedonyolov5s
AT xialiu detectionmethodofpotatoleafdiseasebasedonyolov5s
AT changyiwang detectionmethodofpotatoleafdiseasebasedonyolov5s