针对玉米叶片病害的半监督式单阶段目标检测

Q2 Agricultural and Biological Sciences Agriculture Pub Date : 2024-07-14 DOI:10.3390/agriculture14071140
Jiaqi Liu, Yanxin Hu, Qianfu Su, Jianwei Guo, Zhiyu Chen, Gang Liu
{"title":"针对玉米叶片病害的半监督式单阶段目标检测","authors":"Jiaqi Liu, Yanxin Hu, Qianfu Su, Jianwei Guo, Zhiyu Chen, Gang Liu","doi":"10.3390/agriculture14071140","DOIUrl":null,"url":null,"abstract":"Maize is one of the most important crops globally, and accurate diagnosis of leaf diseases is crucial for ensuring increased yields. Despite the continuous progress in computer vision technology, detecting maize leaf diseases based on deep learning still relies on a large amount of manually labeled data, and the labeling process is time-consuming and labor-intensive. Moreover, the detectors currently used for identifying maize leaf diseases have relatively low accuracy in complex experimental fields. Therefore, the proposed Agronomic Teacher, an object detection algorithm that utilizes limited labeled and abundant unlabeled data, is applied to maize leaf disease recognition. In this work, a semi-supervised object detection framework is built based on a single-stage detector, integrating the Weighted Average Pseudo-labeling Assignment (WAP) strategy and AgroYOLO detector combining Agro-Backbone network with Agro-Neck network. The WAP strategy uses weight adjustments to set objectness and classification scores as evaluation criteria for pseudo-labels reliability assignment. Agro-Backbone network accurately extracts features of maize leaf diseases and obtains richer semantic information. Agro-Neck network enhances feature fusion by utilizing multi-layer features for collaborative combinations. The effectiveness of the proposed method is validated on the MaizeData and PascalVOC datasets at different annotation ratios. Compared to the baseline model, Agronomic Teacher leverages abundant unlabeled data to achieve a 6.5% increase in mAP (0.5) on the 30% labeled MaizeData. On the 30% labeled PascalVOC dataset, the mAP (0.5) improved by 8.2%, demonstrating the method’s potential for generalization.","PeriodicalId":7447,"journal":{"name":"Agriculture","volume":"44 21","pages":""},"PeriodicalIF":0.0000,"publicationDate":"2024-07-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Semi-Supervised One-Stage Object Detection for Maize Leaf Disease\",\"authors\":\"Jiaqi Liu, Yanxin Hu, Qianfu Su, Jianwei Guo, Zhiyu Chen, Gang Liu\",\"doi\":\"10.3390/agriculture14071140\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Maize is one of the most important crops globally, and accurate diagnosis of leaf diseases is crucial for ensuring increased yields. Despite the continuous progress in computer vision technology, detecting maize leaf diseases based on deep learning still relies on a large amount of manually labeled data, and the labeling process is time-consuming and labor-intensive. Moreover, the detectors currently used for identifying maize leaf diseases have relatively low accuracy in complex experimental fields. Therefore, the proposed Agronomic Teacher, an object detection algorithm that utilizes limited labeled and abundant unlabeled data, is applied to maize leaf disease recognition. In this work, a semi-supervised object detection framework is built based on a single-stage detector, integrating the Weighted Average Pseudo-labeling Assignment (WAP) strategy and AgroYOLO detector combining Agro-Backbone network with Agro-Neck network. The WAP strategy uses weight adjustments to set objectness and classification scores as evaluation criteria for pseudo-labels reliability assignment. Agro-Backbone network accurately extracts features of maize leaf diseases and obtains richer semantic information. Agro-Neck network enhances feature fusion by utilizing multi-layer features for collaborative combinations. The effectiveness of the proposed method is validated on the MaizeData and PascalVOC datasets at different annotation ratios. Compared to the baseline model, Agronomic Teacher leverages abundant unlabeled data to achieve a 6.5% increase in mAP (0.5) on the 30% labeled MaizeData. On the 30% labeled PascalVOC dataset, the mAP (0.5) improved by 8.2%, demonstrating the method’s potential for generalization.\",\"PeriodicalId\":7447,\"journal\":{\"name\":\"Agriculture\",\"volume\":\"44 21\",\"pages\":\"\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2024-07-14\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Agriculture\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.3390/agriculture14071140\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q2\",\"JCRName\":\"Agricultural and Biological Sciences\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Agriculture","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.3390/agriculture14071140","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"Agricultural and Biological Sciences","Score":null,"Total":0}
引用次数: 0

摘要

玉米是全球最重要的农作物之一,准确诊断叶片病害对确保增产至关重要。尽管计算机视觉技术不断进步,但基于深度学习的玉米叶病检测仍然依赖于大量人工标注的数据,而且标注过程耗时耗力。此外,目前用于识别玉米叶片病害的检测器在复杂的实验田中准确率相对较低。因此,所提出的农艺老师(Agronomic Teacher)是一种利用有限的标注数据和丰富的非标注数据的对象检测算法,被应用于玉米叶病识别。这项工作基于单级检测器,整合了加权平均伪标记分配(WAP)策略和结合了 Agro-Backbone 网络和 Agro-Neck 网络的 AgroYOLO 检测器,建立了一个半监督对象检测框架。WAP 策略利用权重调整将对象性和分类分数设定为伪标签可靠性分配的评估标准。Agro-Backbone 网络能准确提取玉米叶片病害的特征,并获得更丰富的语义信息。Agro-Neck 网络利用多层特征进行协作组合,增强了特征融合。在不同注释比例的 MaizeData 和 PascalVOC 数据集上验证了所提方法的有效性。与基线模型相比,Agronomic Teacher 利用丰富的未标注数据,在标注率为 30% 的 MaizeData 数据集上,mAP (0.5) 提高了 6.5%。在标注率为 30% 的 PascalVOC 数据集上,mAP (0.5) 提高了 8.2%,证明了该方法的泛化潜力。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
查看原文
分享 分享
微信好友 朋友圈 QQ好友 复制链接
本刊更多论文
Semi-Supervised One-Stage Object Detection for Maize Leaf Disease
Maize is one of the most important crops globally, and accurate diagnosis of leaf diseases is crucial for ensuring increased yields. Despite the continuous progress in computer vision technology, detecting maize leaf diseases based on deep learning still relies on a large amount of manually labeled data, and the labeling process is time-consuming and labor-intensive. Moreover, the detectors currently used for identifying maize leaf diseases have relatively low accuracy in complex experimental fields. Therefore, the proposed Agronomic Teacher, an object detection algorithm that utilizes limited labeled and abundant unlabeled data, is applied to maize leaf disease recognition. In this work, a semi-supervised object detection framework is built based on a single-stage detector, integrating the Weighted Average Pseudo-labeling Assignment (WAP) strategy and AgroYOLO detector combining Agro-Backbone network with Agro-Neck network. The WAP strategy uses weight adjustments to set objectness and classification scores as evaluation criteria for pseudo-labels reliability assignment. Agro-Backbone network accurately extracts features of maize leaf diseases and obtains richer semantic information. Agro-Neck network enhances feature fusion by utilizing multi-layer features for collaborative combinations. The effectiveness of the proposed method is validated on the MaizeData and PascalVOC datasets at different annotation ratios. Compared to the baseline model, Agronomic Teacher leverages abundant unlabeled data to achieve a 6.5% increase in mAP (0.5) on the 30% labeled MaizeData. On the 30% labeled PascalVOC dataset, the mAP (0.5) improved by 8.2%, demonstrating the method’s potential for generalization.
求助全文
通过发布文献求助,成功后即可免费获取论文全文。 去求助
来源期刊
Agriculture
Agriculture Agricultural and Biological Sciences-Horticulture
CiteScore
1.90
自引率
0.00%
发文量
4
审稿时长
11 weeks
期刊介绍: The Agriculture (Poľnohospodárstvo) is a peer-reviewed international journal that publishes mainly original research papers. The journal examines various aspects of research and is devoted to the publication of papers dealing with the following subjects: plant nutrition, protection, breeding, genetics and biotechnology, quality of plant products, grassland, mountain agriculture and environment, soil science and conservation, mechanization and economics of plant production and other spheres of plant science. Journal is published 4 times per year.
期刊最新文献
Effects of Abscisic Acid on Rice Seed Dormancy: Antioxidant Response and Accumulations of Melatonin, Phenolics and Momilactones Classification of Degradable Mulch Films and Their Promotional Effects and Limitations on Agricultural Production Deep Learning with a Multi-Task Convolutional Neural Network to Generate a National-Scale 3D Soil Data Product: The Particle Size Distribution of the German Agricultural Soil Landscape Enhanced Food-Production Efficiencies through Integrated Farming Systems in the Hau Giang Province in the Mekong Delta, Vietnam The Influence of Nitrogen and Sulfur Fertilization on Oil Quality and Seed Meal in Different Genotypes of Winter Oilseed Rape (Brassica napus L.)
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
现在去查看 取消
×
提示
确定
0
微信
客服QQ
Book学术公众号 扫码关注我们
反馈
×
意见反馈
请填写您的意见或建议
请填写您的手机或邮箱
已复制链接
已复制链接
快去分享给好友吧!
我知道了
×
扫码分享
扫码分享
Book学术官方微信
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术
文献互助 智能选刊 最新文献 互助须知 联系我们:info@booksci.cn
Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。
Copyright © 2023 Book学术 All rights reserved.
ghs 京公网安备 11010802042870号 京ICP备2023020795号-1