Article(id=1256183371516993788, tenantId=1146029695717560320, journalId=1255847919539208197, issueId=1256183358493679805, articleNumber=null, orderNo=null, doi=10.13193/j.issn.1673-7717.2025.12.004, pmid=null, cstr=null, oa=null, hot=null, price=null, onlineType=0, articleFormat=0, articleType=null, articleTypeStr=null, receivedDate=null, receivedDateStr=null, revisedDate=null, revisedDateStr=null, acceptedDate=null, acceptedDateStr=null, onlineDate=1777427054528, onlineDateStr=2026-04-29, pubDate=1765296000000, pubDateStr=2025-12-10, doiRegisterDate=null, doiRegisterDateStr=null, onlineIssueDate=1777427054528, onlineIssueDateStr=2026-04-29, onlineJustAcceptDate=null, onlineJustAcceptDateStr=null, onlineFirstDate=null, onlineFirstDateStr=null, sourceXml=null, magXml=null, createTime=1777427054528, creator=13701087609, updateTime=1777427054528, updator=13701087609, issue=Issue{id=1256183358493679805, tenantId=1146029695717560320, journalId=1255847919539208197, year='2025', volume='43', issue='12', pageStart='1', pageEnd='258', issueExtLink='null', onlineDate='null', pubDate='null', beforeIssueId=null, nextIssueId=null, price=null, status=1, issueComplete=1, articleOrder=1, issueType=1, specialIssue=null, createTime=1777427051344, creator=13701087609, updateTime=1777427760067, updator=13701087609, preIssue=null, nextIssue=null, ext={EN=IssueExt(id=1256186331126969089, tenantId=1146029695717560320, journalId=1255847919539208197, issueId=1256183358493679805, language=EN, specialIssueTitle=, coverIllustrator=null, specialIssueEditor=, specialIssueAbout=), CN=IssueExt(id=1256186331126969090, tenantId=1146029695717560320, journalId=1255847919539208197, issueId=1256183358493679805, language=CN, specialIssueTitle=, coverIllustrator=null, specialIssueEditor=, specialIssueAbout=)}, issueFiles=null}, startPage=21, endPage=25, ext={EN=ArticleExt(id=1256183377330299180, articleId=1256183371516993788, tenantId=1146029695717560320, journalId=1255847919539208197, language=EN, title=Research on Mobile Oriented Lightweight YOLOv10 Algorithm for Chinese Herbal Medicine Detection, columnId=1256183375241535778, journalTitle=Chinese Archives of Traditional Chinese Medicine, columnName=Digital Traditional Chinese Medicine, runingTitle=null, highlight=null, articleAbstract=

In this research,the lightweight deep learning YOLOv10 object detection algorithm is harnessed to develop an efficient algorithm for detecting Chinese herbal medicines on mobile devices,furnishing an intelligent and high-performance technical solution for the detection of Chinese herbal medicines in scenarios such as cultivation production,quality assessment,educational popularization and automated dispensing.A dataset for target detection of 31 commonly used Chinese herbal medicines is constructed,comprising a total of 6900 images.To enhance the multi-scale detection capability and detection efficiency of the model,the Ghost Convolution(Ghost Convolution,Ghost Conv)lightweight convolution and the weighted bidirectional feature pyramid module(Bidirectional Feature Pyramid Network,BiFPN)are incorporated,and an improved lightweight model for Chinese herbal medicine detection,namely YOLOv10n-GB,is proposed.The algorithm is then employed to conduct training,testing,analysis and mobile terminal deployment tests on the image samples within the dataset.The improved model exhibits2.27 M(Million)parameters,6.4 G(Giga)computational complexity,and an m AP50 value of0.947.In comparison with YOLOv10n,the parameter count is reduced by 1.2%,the computational load is decreased by 4.5%,and the mean average precision 50(mAP50)is augmented by 1.8%.The average frame rate of the model detection on the mobile terminal reaches 8.1Frames Per Second(FPS).When contrasted with other lightweight algorithms such as YOLOv5n and YOLOv8n,YOLOv10n-GB demonstrates the lowest floating-point computational cost and higher detection accuracy.The YOLOv10n-GB algorithm has accomplished the task of efficient detection of Chinese herbal medicines on mobile terminals,paving the way for novel ideas in establishing a portable and real-time detection scheme for Chinese herbal medicines.

, correspAuthors=null, authorNote=null, correspAuthorsNote=null, copyrightStatement=null, copyrightOwner=null, extLink=null, articleAbsUrl=null, sourceXml=null, magXml=null, pdfUrl=null, pdf=null, pdfFileSize=null, pdfExtLink=null, richHtmlUrl=null, mobilePdfUrl=null, reviewReport=null, pdfFirstPage=null, abstractGraph=null, abstractGraphContent=null, abstractVideo=null, citation=null, cebUrl=null, magXmlContent=null, mapNumber=null, authorCompany=null, fund=null, authors=null, authorsList=Weijie ZHOU, Lijian PANG, Xiaodong LYU, Yanjie CHENG, Xiaowen XIE, Jinge QIU), CN=ArticleExt(id=1256183517839482933, articleId=1256183371516993788, tenantId=1146029695717560320, journalId=1255847919539208197, language=CN, title=面向移动端的轻量化YOLOv10中药材检测算法研究, columnId=1256183387140776248, journalTitle=中华中医药学刊, columnName=数字中医中药, runingTitle=null, highlight=null, articleAbstract=

利用轻量化的深度学习YOLO(You Only Look Once,YOLO)v10目标检测算法,构建了面向移动端的高效中药材检测算法,为中药材种植生产、质量评价、教学科普以及自动化抓药等工作场景中的中药材检测提供了智能、高效的技术方案。建立包含31种常用中药材目标检测数据集,共6900张图像。为提升模型的多尺度检测能力和检测效率,引入幻影卷积(Ghost Convolution,GhostConv)轻量化卷积与加权双向特征金字塔模块(Bidirectional Feature Pyramid Network,BiFPN),提出了一种改进的轻量化中药材检测模型YOLOv10n-GB,并利用算法对数据集中图像样本进行训练、测试、分析、移动端部署测试。改进后模型的参数量2.27M(Million,M)、计算量6.4G(Giga,G)、平均精度mAP50(mean Average Precision 50,mAP50)为0.947,与YOLOv10n相比参数量降低1.2%、计算量降低4.5%,mAP50提升了1.8%,在移动端的模型检测平均帧率8.1帧每秒(Frames Per Second,FPS)。与其他轻量化算法YOLOv5n、YOLOv8n相比,YOLOv10n-GB具有最低的浮点计算量和更高的检测精度。YOLOv10n-GB算法完成了面向移动端的中药材的高效检测任务,为建立便携、实时的中药材检测方案开辟了新思路。

, correspAuthors=null, authorNote=null, correspAuthorsNote=
吕晓东(1966-),女,辽宁沈阳人,教授,博士研究生导师,博士,研究方向:中医肺病与络病基础与临床。
, copyrightStatement=null, copyrightOwner=null, extLink=null, articleAbsUrl=null, sourceXml=D/AVOL/luU2Pj8U9lWNvUw==, magXml=8Qzc6PdcKqVb6QGnr9mafg==, pdfUrl=null, pdf=W387LymSRr5YN9JuHeiv7A==, pdfFileSize=3280643, pdfExtLink=null, richHtmlUrl=null, mobilePdfUrl=null, reviewReport=null, pdfFirstPage=null, abstractGraph=6kLXbBdhtD/z3YKMNxQSqw==, abstractGraphContent=null, abstractVideo=null, citation=null, cebUrl=null, magXmlContent=eJiHT7XhqnD+LXj7Ji5bTA==, mapNumber=null, authorCompany=null, fund=null, authors=

周伟杰(1994-),男,辽宁大连人,助教,硕士,研究方向:中医人工智能、机器视觉、中医大数据分析。

, authorsList=周伟杰, 庞立健, 吕晓东, 程彦杰, 谢晓雯, 邱巾格)}, authors=[Author(id=1256183528354603158, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, orderNo=0, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1256183530422395049, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, authorId=1256183528354603158, language=EN, stringName=Weijie ZHOU, firstName=Weijie, middleName=null, lastName=ZHOU, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1.Liaoning University of Traditional Chinese Medicine,Shenyang 110847,Liaoning,China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1256183531751989428, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, authorId=1256183528354603158, language=CN, stringName=周伟杰, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1.辽宁中医药大学,辽宁 沈阳 110847, bio={"content":"

周伟杰(1994-),男,辽宁大连人,助教,硕士,研究方向:中医人工智能、机器视觉、中医大数据分析。

"}, bioImg=null, bioContent=

周伟杰(1994-),男,辽宁大连人,助教,硕士,研究方向:中医人工智能、机器视觉、中医大数据分析。

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1256183525200486513, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, xref=1., ext=[AuthorCompanyExt(id=1256183525406007411, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, companyId=1256183525200486513, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1.Liaoning University of Traditional Chinese Medicine,Shenyang 110847,Liaoning,China), AuthorCompanyExt(id=1256183525812854901, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, companyId=1256183525200486513, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1.辽宁中医药大学,辽宁 沈阳 110847)])]), Author(id=1256183533882695882, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, orderNo=1, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1256183534755111128, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, authorId=1256183533882695882, language=EN, stringName=Lijian PANG, firstName=Lijian, middleName=null, lastName=PANG, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=2, address=2.Liaoning University of Traditional Chinese Medicine Affiliated Hospital,Shenyang 110032,Liaoning,China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1256183535954682077, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, authorId=1256183533882695882, language=CN, stringName=庞立健, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=2, address=2.辽宁中医药大学附属医院,辽宁 沈阳 110032, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1256183526710435970, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, xref=2., ext=[AuthorCompanyExt(id=1256183527062757507, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, companyId=1256183526710435970, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2.Liaoning University of Traditional Chinese Medicine Affiliated Hospital,Shenyang 110032,Liaoning,China), AuthorCompanyExt(id=1256183527104700551, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, companyId=1256183526710435970, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2.辽宁中医药大学附属医院,辽宁 沈阳 110032)])]), Author(id=1256183536726434023, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, orderNo=2, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1256183538441904379, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, authorId=1256183536726434023, language=EN, stringName=Xiaodong LYU, firstName=Xiaodong, middleName=null, lastName=LYU, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1.Liaoning University of Traditional Chinese Medicine,Shenyang 110847,Liaoning,China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1256183539779887362, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, authorId=1256183536726434023, language=CN, stringName=吕晓东, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1.辽宁中医药大学,辽宁 沈阳 110847, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1256183525200486513, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, xref=1., ext=[AuthorCompanyExt(id=1256183525406007411, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, companyId=1256183525200486513, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1.Liaoning University of Traditional Chinese Medicine,Shenyang 110847,Liaoning,China), AuthorCompanyExt(id=1256183525812854901, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, companyId=1256183525200486513, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1.辽宁中医药大学,辽宁 沈阳 110847)])]), Author(id=1256183541633769746, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, orderNo=3, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1256183542736871707, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, authorId=1256183541633769746, language=EN, stringName=Yanjie CHENG, firstName=Yanjie, middleName=null, lastName=CHENG, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1.Liaoning University of Traditional Chinese Medicine,Shenyang 110847,Liaoning,China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1256183543462486310, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, authorId=1256183541633769746, language=CN, stringName=程彦杰, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1.辽宁中医药大学,辽宁 沈阳 110847, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1256183525200486513, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, xref=1., ext=[AuthorCompanyExt(id=1256183525406007411, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, companyId=1256183525200486513, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1.Liaoning University of Traditional Chinese Medicine,Shenyang 110847,Liaoning,China), AuthorCompanyExt(id=1256183525812854901, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, companyId=1256183525200486513, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1.辽宁中医药大学,辽宁 沈阳 110847)])]), Author(id=1256183544011940145, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, orderNo=4, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1256183545119236415, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, authorId=1256183544011940145, language=EN, stringName=Xiaowen XIE, firstName=Xiaowen, middleName=null, lastName=XIE, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1.Liaoning University of Traditional Chinese Medicine,Shenyang 110847,Liaoning,China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1256183545958097219, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, authorId=1256183544011940145, language=CN, stringName=谢晓雯, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1.辽宁中医药大学,辽宁 沈阳 110847, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1256183525200486513, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, xref=1., ext=[AuthorCompanyExt(id=1256183525406007411, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, companyId=1256183525200486513, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1.Liaoning University of Traditional Chinese Medicine,Shenyang 110847,Liaoning,China), AuthorCompanyExt(id=1256183525812854901, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, companyId=1256183525200486513, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1.辽宁中医药大学,辽宁 沈阳 110847)])]), Author(id=1256183546780180812, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, orderNo=5, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1256183547707122002, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, authorId=1256183546780180812, language=EN, stringName=Jinge QIU, firstName=Jinge, middleName=null, lastName=QIU, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1.Liaoning University of Traditional Chinese Medicine,Shenyang 110847,Liaoning,China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1256183548927664478, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, authorId=1256183546780180812, language=CN, stringName=邱巾格, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1.辽宁中医药大学,辽宁 沈阳 110847, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1256183525200486513, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, xref=1., ext=[AuthorCompanyExt(id=1256183525406007411, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, companyId=1256183525200486513, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1.Liaoning University of Traditional Chinese Medicine,Shenyang 110847,Liaoning,China), AuthorCompanyExt(id=1256183525812854901, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, companyId=1256183525200486513, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1.辽宁中医药大学,辽宁 沈阳 110847)])])], keywords=[Keyword(id=1256183551037399415, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, language=EN, orderNo=1, keyword=deep learning), Keyword(id=1256183552325050755, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, language=EN, orderNo=2, keyword=object detection), Keyword(id=1256183553524621708, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, language=EN, orderNo=3, keyword=YOLOv10), Keyword(id=1256183554795495832, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, language=EN, orderNo=4, keyword=BiFPN), Keyword(id=1256183556196393396, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, language=EN, orderNo=5, keyword=GhostConv), Keyword(id=1256183556980728260, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, language=EN, orderNo=6, keyword=Chinese herbal medicine detection), Keyword(id=1256183557790228948, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, language=CN, orderNo=1, keyword=深度学习), Keyword(id=1256183558616506848, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, language=CN, orderNo=2, keyword=目标检测), Keyword(id=1256183559862215149, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, language=CN, orderNo=3, keyword=YOLOv10), Keyword(id=1256183560747213302, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, language=CN, orderNo=4, keyword=BiFPN), Keyword(id=1256183561552519681, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, language=CN, orderNo=5, keyword=GhostConv), Keyword(id=1256183562458489356, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, language=CN, orderNo=6, keyword=中药材检测)], refs=[Reference(id=1256183579923571431, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, doi=null, pmid=null, pmcid=null, year=2003, volume=null, issue=4, pageStart=4, pageEnd=7, url=null, language=null, rfNumber=[1], rfOrder=0, authorNames=冒巍巍, 范莉, 王方华, journalName=中草药, refType=null, unstructuredReference=冒巍巍,范莉,王方华. 我国中药产业国际竞争力分析[J]. 中草药2003(4):4-7., articleTitle=我国中药产业国际竞争力分析, refAbstract=null), Reference(id=1256183580431082222, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, doi=null, pmid=null, pmcid=null, year=2022, volume=40, issue=8, pageStart=47, pageEnd=50, url=null, language=null, rfNumber=[2], rfOrder=1, authorNames=徐雅静, 俞捷, 余远盼, journalName=中华中医药学刊, refType=null, unstructuredReference=徐雅静,俞捷,余远盼,. 人工智能在中药材及饮片鉴别领域的应用[J]. 中华中医药学刊202240(8):47-50., articleTitle=人工智能在中药材及饮片鉴别领域的应用, refAbstract=null), Reference(id=1256183580829541105, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, doi=null, pmid=null, pmcid=null, year=2023, volume=18, issue=6, pageStart=42, pageEnd=50, url=null, language=null, rfNumber=[3], rfOrder=2, authorNames=谭代庆, 肖志鸿, 吴浩忠, journalName=中国数字医学, refType=null, unstructuredReference=谭代庆,肖志鸿,吴浩忠,. 融合改进残差网络和Transformer的易混淆中药饮片识别研究[J]. 中国数字医学202318(6):42-50., articleTitle=融合改进残差网络和Transformer的易混淆中药饮片识别研究, refAbstract=null), Reference(id=1256183581097976566, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, doi=null, pmid=null, pmcid=null, year=2024, volume=47, issue=3, pageStart=29, pageEnd=33, url=null, language=null, rfNumber=[4], rfOrder=3, authorNames=张琦, 区锦锋, 周华英, journalName=现代电子技术, refType=null, unstructuredReference=张琦,区锦锋,周华英. 基于Xception与迁移学习的中药饮片图像识别研究[J]. 现代电子技术202447(3):29-33., articleTitle=基于Xception与迁移学习的中药饮片图像识别研究, refAbstract=null), Reference(id=1256183581316080378, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, doi=null, pmid=null, pmcid=null, year=2024, volume=24, issue=9, pageStart=3596, pageEnd=3604, url=null, language=null, rfNumber=[5], rfOrder=4, authorNames=李玥辰, 赵晓, 王若男, journalName=科学技术与工程, refType=null, unstructuredReference=李玥辰,赵晓,王若男,. 基于改进卷积神经网络的中药饮片图像识别[J]. 科学技术与工程202424(9):3596-3604., articleTitle=基于改进卷积神经网络的中药饮片图像识别, refAbstract=null), Reference(id=1256183581693567745, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, doi=null, pmid=null, pmcid=null, year=2023, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[6], rfOrder=5, authorNames=周丽媛, journalName=null, refType=null, unstructuredReference=周丽媛. 基于深度学习的草本藏药材识别方法研究[D]. 拉萨:西藏大学,2023., articleTitle=基于深度学习的草本藏药材识别方法研究, refAbstract=null), Reference(id=1256183581794231047, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, doi=null, pmid=null, pmcid=null, year=2024, volume=43, issue=9, pageStart=130, pageEnd=140, url=null, language=null, rfNumber=[7], rfOrder=6, authorNames=谷瑞, 宋翠玲, 李元昊, journalName=国外电子测量技术, refType=null, unstructuredReference=谷瑞,宋翠玲,李元昊. 改进Res2Net和注意力的中药饮片识别模型[J]. 国外电子测量技术202443(9):130-140., articleTitle=改进Res2Net和注意力的中药饮片识别模型, refAbstract=null), Reference(id=1256183582196884238, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, doi=null, pmid=null, pmcid=null, year=2023, volume=46, issue=1, pageStart=57, pageEnd=61, url=null, language=null, rfNumber=[8], rfOrder=7, authorNames=高爽, 周志强, 黄显章, journalName=中药材, refType=null, unstructuredReference=高爽,周志强,黄显章,. 基于YOLOv3算法的中药饮片检测研究[J]. 中药材202346(1):57-61., articleTitle=基于YOLOv3算法的中药饮片检测研究, refAbstract=null), Reference(id=1256183582444348176, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, doi=null, pmid=null, pmcid=null, year=2023, volume=29, issue=14, pageStart=133, pageEnd=140, url=null, language=null, rfNumber=[9], rfOrder=8, authorNames=郭丛, 田钰嘉, 李杨, journalName=中国实验方剂学杂志, refType=null, unstructuredReference=郭丛,田钰嘉,李杨,. 基于YOLOv4算法的中药饮片识别[J]. 中国实验方剂学杂志202329(14):133-140., articleTitle=基于YOLOv4算法的中药饮片识别, refAbstract=null), Reference(id=1256183582540817171, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, doi=null, pmid=null, pmcid=null, year=2024, volume=35, issue=12, pageStart=2898, pageEnd=2904, url=null, language=null, rfNumber=[10], rfOrder=9, authorNames=华畅, 郑豪, journalName=时珍国医国药, refType=null, unstructuredReference=华畅,郑豪. 面向中药饮片识别的MSDA-YOLOv8检测算法[J]. 时珍国医国药202435(12):2898-2904., articleTitle=面向中药饮片识别的MSDA-YOLOv8检测算法, refAbstract=null), Reference(id=1256183582641480470, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, doi=null, pmid=null, pmcid=null, year=2024, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[11], rfOrder=10, authorNames=VARGHESE R, SAMBATH M, journalName=null, refType=null, unstructuredReference=VARGHESE R,SAMBATH M. YOLOv8:A Novel Object Detection Algorithm with Enhanced Performance and Robustness[C]//2024 International Conference on Advances in Data Engineering and Intelligent Computing Systems(ADICS),2024., articleTitle=YOLOv8:A Novel Object Detection Algorithm with Enhanced Performance and Robustness, refAbstract=null), Reference(id=1256183582863778584, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, doi=null, pmid=null, pmcid=null, year=2016, volume=null, issue=null, pageStart=770, pageEnd=778, url=null, language=null, rfNumber=[12], rfOrder=11, authorNames=HE K, ZHANG X, REN S, journalName=Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition(CVPR), refType=null, unstructuredReference=HE K,ZHANG X,REN S,et al. Deep residual learning for image recognition[J]. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition(CVPR)2016:770-778., articleTitle=Deep residual learning for image recognition, refAbstract=null), Reference(id=1256183582993802014, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, doi=null, pmid=null, pmcid=null, year=2023, volume=45, issue=6, pageStart=10, pageEnd=20,26, url=null, language=null, rfNumber=[13], rfOrder=12, authorNames=郭庆梅, 刘宁波, 王中训, journalName=探测与控制学报, refType=null, unstructuredReference=郭庆梅,刘宁波,王中训,. 基于深度学习的目标检测算法综述[J]. 探测与控制学报202345(6):10-20,26., articleTitle=基于深度学习的目标检测算法综述, refAbstract=null), Reference(id=1256183583304180513, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, doi=null, pmid=null, pmcid=null, year=2022, volume=16, issue=5, pageStart=1025, pageEnd=1042, url=null, language=null, rfNumber=[14], rfOrder=13, authorNames=董文轩, 梁宏涛, 刘国柱, journalName=计算机科学与探索, refType=null, unstructuredReference=董文轩,梁宏涛,刘国柱,. 深度卷积应用于目标检测算法综述[J]. 计算机科学与探索202216(5):1025-1042., articleTitle=深度卷积应用于目标检测算法综述, refAbstract=null), Reference(id=1256183583870411555, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, doi=null, pmid=null, pmcid=null, year=2024, volume=10,30, issue=1, pageStart=1, pageEnd=8, url=null, language=null, rfNumber=[15], rfOrder=14, authorNames=黄毅, 周纯, 刘欣军, journalName=光通信研究, refType=null, unstructuredReference=黄毅,周纯,刘欣军,. 基于YOLOv10的无人机复杂背景下多尺度检测模型[J]. 光通信研究202410,30(1):1-8., articleTitle=基于YOLOv10的无人机复杂背景下多尺度检测模型, refAbstract=null), Reference(id=1256183584126264102, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, doi=null, pmid=null, pmcid=null, year=2015, volume=null, issue=null, pageStart=1440, pageEnd=1448, url=null, language=null, rfNumber=[16], rfOrder=15, authorNames=GIRSHICK R, journalName=null, refType=null, unstructuredReference=GIRSHICK R. Fast R-CNN[C]//International Conference on Computer Vision. IEEE Computer Society,2015:1440-1448., articleTitle=Fast R-CNN, refAbstract=null), Reference(id=1256183584252093225, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, doi=null, pmid=null, pmcid=null, year=2017, volume=39, issue=6, pageStart=1137, pageEnd=1149, url=null, language=null, rfNumber=[17], rfOrder=16, authorNames=REN S, HE K, GIRSHICK R, journalName=IEEE Transactions on Pattern Analysis & Machine Intelligence, refType=null, unstructuredReference=REN S,HE K,GIRSHICK R,et al. Faster R-CNN:towards realtime object detection with region proposal networks[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence201739(6):1137-1149., articleTitle=Faster R-CNN:towards realtime object detection with region proposal networks, refAbstract=null), Reference(id=1256183584575054637, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, doi=null, pmid=null, pmcid=null, year=2017, volume=null, issue=null, pageStart=2961, pageEnd=2969, url=null, language=null, rfNumber=[18], rfOrder=17, authorNames=HE K M, GKIOXARIG, PIOTR D, journalName=null, refType=null, unstructuredReference=HE K M,GKIOXARIG,PIOTR D,et al. MaskR-CNN[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition(CVPR),2017:2961-2969., articleTitle=MaskR-CNN, refAbstract=null), Reference(id=1256183584948347696, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, doi=null, pmid=null, pmcid=null, year=2019, volume=null, issue=null, pageStart=821, pageEnd=830, url=null, language=null, rfNumber=[19], rfOrder=18, authorNames=PANG J, CHEN K, SHI J, journalName=null, refType=null, unstructuredReference=PANG J,CHEN K,SHI J,et al. Libra R-CNN:towards balanced learning for object detection[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition,2019:821-830., articleTitle=Libra R-CNN:towards balanced learning for object detection, refAbstract=null), Reference(id=1256183585036428082, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, doi=null, pmid=null, pmcid=null, year=2019, volume=null, issue=null, pageStart=7363, pageEnd=7372, url=null, language=null, rfNumber=[20], rfOrder=19, authorNames=LU X, LIB, YU EY, journalName=null, refType=null, unstructuredReference=LU X,LIB,YU EY,et al. GridR-CNN[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition(CVPR),2019:7363-7372., articleTitle=GridR-CNN, refAbstract=null), Reference(id=1256183585145479989, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, doi=null, pmid=null, pmcid=null, year=2016, volume=null, issue=null, pageStart=21, pageEnd=37, url=null, language=null, rfNumber=[21], rfOrder=20, authorNames=LIUW, ANGUELOV D, ERHAN D, journalName=null, refType=null, unstructuredReference=LIUW,ANGUELOV D,ERHAN D,et al. SSD:Single Shot Multi-Box Detector[C]//Proceedings of the Computer Vision(CVPR)-ECCV 2016. Springer International Publishing,2016:21-37., articleTitle=SSD:Single Shot Multi-Box Detector, refAbstract=null), Reference(id=1256183585388749622, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, doi=null, pmid=null, pmcid=null, year=2023, volume=null, issue=null, pageStart=7464, pageEnd=7475, url=null, language=null, rfNumber=[22], rfOrder=21, authorNames=WANG C Y, BOCHKOVSKIY A, LIAO H Y M, journalName=null, refType=null, unstructuredReference=WANG C Y,BOCHKOVSKIY A,LIAO H Y M. YOLOv7:Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors[C]//IEEE/CVF Conference on Computer Vision and Pattern Recognition(CVPR),Vancouver,Canada. IEEE,2023:7464-7475., articleTitle=YOLOv7:Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors, refAbstract=null), Reference(id=1256183585581687610, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, doi=null, pmid=null, pmcid=null, year=2024, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[23], rfOrder=22, authorNames=WANG C Y, YEH IH, LIAO H Y M, journalName=null, refType=null, unstructuredReference=WANG C Y,YEH IH,LIAO H Y M. Yolov9:Learning what you want to learn using programmable gradient information[C]//IEEE Conference on Computer Vision and Pattern Recognition(CVPR). SeattleWA,USA:IEEE,2024., articleTitle=Yolov9:Learning what you want to learn using programmable gradient information, refAbstract=null), Reference(id=1256183586227610431, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, doi=null, pmid=null, pmcid=null, year=2025, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[24], rfOrder=23, authorNames=WANG A, CHEN H, LIU L, journalName=null, refType=null, unstructuredReference=WANG A,CHEN H,LIU L,et al. Yolov10:Real-time end-toend object detection[C]//IEEE Conference on Computer Vision and Pattern Recognition(CVPR). SeattleWA,USA:IEEE,2025., articleTitle=Yolov10:Real-time end-toend object detection, refAbstract=null), Reference(id=1256183586374411073, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, doi=null, pmid=null, pmcid=null, year=2024, volume=null, issue=1, pageStart=1, pageEnd=13, url=null, language=null, rfNumber=[25], rfOrder=24, authorNames=高立鹏, 周孟然, 胡锋, journalName=煤炭科学技术, refType=null, unstructuredReference=高立鹏,周孟然,胡锋,. 基于REIW-YOLOv10n的井下安全帽小目标检测算法[J]. 煤炭科学技术2024(1):1-13., articleTitle=基于REIW-YOLOv10n的井下安全帽小目标检测算法, refAbstract=null), Reference(id=1256183586756092739, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, doi=null, pmid=null, pmcid=null, year=2020, volume=null, issue=null, pageStart=1577, pageEnd=1586, url=null, language=null, rfNumber=[26], rfOrder=25, authorNames=K HAN, Y WANG, Q TIAN, journalName=null, refType=null, unstructuredReference=K HAN,Y WANG,Q TIAN,et al. GhostNet:More Features From Cheap Operations[C]//2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition(CVPR),Seattle,WA,USA,2020:1577-1586., articleTitle=GhostNet:More Features From Cheap Operations, refAbstract=null), Reference(id=1256183587087442757, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, doi=null, pmid=null, pmcid=null, year=2020, volume=null, issue=null, pageStart=10778, pageEnd=10787, url=null, language=null, rfNumber=[27], rfOrder=26, authorNames=TAN M M, PANG R, LE Q V, journalName=null, refType=null, unstructuredReference=TAN M M,PANG R,LE Q V. EfficientDet:Scalable and Efficient Object Detection[C]//2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition(CVPR),2020:10778-10787., articleTitle=EfficientDet:Scalable and Efficient Object Detection, refAbstract=null)], funds=[Fund(id=1256183578250044110, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, awardId=82274440, language=CN, fundingSource=国家自然科学基金面上项目(82274440), fundOrder=null, country=null), Fund(id=1256183578417816276, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, awardId=T0302, language=CN, fundingSource=国家中医药管理局中医络病重点学科建设项目(T0302), fundOrder=null, country=null), Fund(id=1256183579126653663, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, awardId=LNZYYCXTD-CCCX-001, language=CN, fundingSource=辽宁省中医药创新团队项目(LNZYYCXTD-CCCX-001), fundOrder=null, country=null)], companyList=[AuthorCompany(id=1256183525200486513, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, xref=1., ext=[AuthorCompanyExt(id=1256183525406007411, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, companyId=1256183525200486513, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1.Liaoning University of Traditional Chinese Medicine,Shenyang 110847,Liaoning,China), AuthorCompanyExt(id=1256183525812854901, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, companyId=1256183525200486513, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1.辽宁中医药大学,辽宁 沈阳 110847)]), AuthorCompany(id=1256183526710435970, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, xref=2., ext=[AuthorCompanyExt(id=1256183527062757507, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, companyId=1256183526710435970, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2.Liaoning University of Traditional Chinese Medicine Affiliated Hospital,Shenyang 110032,Liaoning,China), AuthorCompanyExt(id=1256183527104700551, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, companyId=1256183526710435970, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2.辽宁中医药大学附属医院,辽宁 沈阳 110032)])], figs=[ArticleFig(id=1256183565352559146, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, language=EN, label=null, caption=null, figureFileSmall=Yhx2kfiXQMSJTcLb2P/XpA==, figureFileBig=6kLXbBdhtD/z3YKMNxQSqw==, tableContent=null), ArticleFig(id=1256183566522769972, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, language=CN, label=图1, caption=数据库中药材示例及信息统计, figureFileSmall=Yhx2kfiXQMSJTcLb2P/XpA==, figureFileBig=6kLXbBdhtD/z3YKMNxQSqw==, tableContent=null), ArticleFig(id=1256183569160987215, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, language=EN, label=null, caption=null, figureFileSmall=AMc7yIwA2jiZL1h0opjx/w==, figureFileBig=5/fn0K2tTcjymCVBFZva6Q==, tableContent=null), ArticleFig(id=1256183569932739160, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, language=CN, label=图2, caption=YOLOv10n网络结构, figureFileSmall=AMc7yIwA2jiZL1h0opjx/w==, figureFileBig=5/fn0K2tTcjymCVBFZva6Q==, tableContent=null), ArticleFig(id=1256183570243117658, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, language=EN, label=null, caption=null, figureFileSmall=obmgRbR2q2Bzvco30IFKvQ==, figureFileBig=ZqtTb/5OVvnOwBEMBIJD0w==, tableContent=null), ArticleFig(id=1256183570784182884, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, language=CN, label=图3, caption=GhostConv网络结构, figureFileSmall=obmgRbR2q2Bzvco30IFKvQ==, figureFileBig=ZqtTb/5OVvnOwBEMBIJD0w==, tableContent=null), ArticleFig(id=1256183571287499368, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, language=EN, label=null, caption=null, figureFileSmall=X9JTho2JnAt2ZLhE94x07A==, figureFileBig=B+ZfXu32L7ZtA6a7/2pXiA==, tableContent=null), ArticleFig(id=1256183572369629811, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, language=CN, label=图4, caption=多尺度特征融合图, figureFileSmall=X9JTho2JnAt2ZLhE94x07A==, figureFileBig=B+ZfXu32L7ZtA6a7/2pXiA==, tableContent=null), ArticleFig(id=1256183572793254522, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, language=EN, label=null, caption=null, figureFileSmall=F301Y8cP+XXPJkIAkwG8Rg==, figureFileBig=D1CsYcAFlUfIm4YZ6xVHGQ==, tableContent=null), ArticleFig(id=1256183573401428610, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, language=CN, label=图5, caption=安卓移动终端APP检测效果, figureFileSmall=F301Y8cP+XXPJkIAkwG8Rg==, figureFileBig=D1CsYcAFlUfIm4YZ6xVHGQ==, tableContent=null), ArticleFig(id=1256183574139626123, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, language=EN, label=null, caption=null, figureFileSmall=null, figureFileBig=null, tableContent=
类型代表算法特点
二阶段FastR-CNN[16]、FasterR-CNN[17]、Mask
R-CNN[18]、LibraR-CNN[19]、Grid
R-CNN[20]
利用Region Proposal Network产生多个锚框(anchor box),进行后精细分类,进行非极大值抑制(NMS)后输出最终结果,识别精度高,但是计算量大、识别速度较低。
一阶段SSD[21]、YOLOv7[22]、YOLOv8、YOLOv9[23]、YOLOv10[24]将目标检测任务转化为一个单一的回归问题,卷积神经网络直接在不同尺度的特征图上预测目标位置和类别,无需像两阶段算法先生成候选区域,因此检测速度比二阶段更快,这种统一的检测框架极大地简化了检测流程,提高了检测速度与效率。
), ArticleFig(id=1256183574592610963, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, language=CN, label=表1, caption=

基于深度学习的目标检测算法特点对比

, figureFileSmall=null, figureFileBig=null, tableContent=
类型代表算法特点
二阶段FastR-CNN[16]、FasterR-CNN[17]、Mask
R-CNN[18]、LibraR-CNN[19]、Grid
R-CNN[20]
利用Region Proposal Network产生多个锚框(anchor box),进行后精细分类,进行非极大值抑制(NMS)后输出最终结果,识别精度高,但是计算量大、识别速度较低。
一阶段SSD[21]、YOLOv7[22]、YOLOv8、YOLOv9[23]、YOLOv10[24]将目标检测任务转化为一个单一的回归问题,卷积神经网络直接在不同尺度的特征图上预测目标位置和类别,无需像两阶段算法先生成候选区域,因此检测速度比二阶段更快,这种统一的检测框架极大地简化了检测流程,提高了检测速度与效率。
), ArticleFig(id=1256183574907183768, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, language=EN, label=null, caption=null, figureFileSmall=null, figureFileBig=null, tableContent=
模型GhostConvConcat_bifpnFLOPs/(G)参数量/(MB)mAP50
YOLOv10n6.72.299 2640.929
YOLOv10n-B6.72.299 2760.945
YOLOv10n-G6.42.271 3120.934
YOLOv10n-GB6.42.271 3240.947
), ArticleFig(id=1256183575322419873, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, language=CN, label=表2, caption=

消融实验对比

, figureFileSmall=null, figureFileBig=null, tableContent=
模型GhostConvConcat_bifpnFLOPs/(G)参数量/(MB)mAP50
YOLOv10n6.72.299 2640.929
YOLOv10n-B6.72.299 2760.945
YOLOv10n-G6.42.271 3120.934
YOLOv10n-GB6.42.271 3240.947
), ArticleFig(id=1256183575901233835, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, language=EN, label=null, caption=null, figureFileSmall=null, figureFileBig=null, tableContent=
算法名称PRmAP50参数量/(MB)FLOPs(G)/平均帧率/(FPS)
YOLOv5n0.9310.8760.9342.78.3714
YOLOv8n0.9210.8870.9393.28.7699
YOLOv10n0.917 0.8820.9292.296.7769
YOLOv10n-GB(our)0.932 0.8840.9472.276.4769
), ArticleFig(id=1256183576266138291, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, language=CN, label=表3, caption=

轻量化目标检测算法对比实验

, figureFileSmall=null, figureFileBig=null, tableContent=
算法名称PRmAP50参数量/(MB)FLOPs(G)/平均帧率/(FPS)
YOLOv5n0.9310.8760.9342.78.3714
YOLOv8n0.9210.8870.9393.28.7699
YOLOv10n0.917 0.8820.9292.296.7769
YOLOv10n-GB(our)0.932 0.8840.9472.276.4769
), ArticleFig(id=1256183576710734521, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, language=EN, label=null, caption=null, figureFileSmall=null, figureFileBig=null, tableContent=
类别PRmAP50类别PRmAP50类别PRmAP50
槟榔1.001.001.00钩藤0.980.950.96蝼蛄0.960.940.99
无患子0.991.000.99土元0.900.960.96虻虫0.960.950.99
蟑螂0.980.980.99枳壳0.920.840.95桂皮0.960.960.99
地龙0.921.000.99黄柏0.870.890.95蟋蟀0.981.000.99
僵蝉0.970.970.99枸杞子0.970.880.95草果0.990.800.97
粪金龟0.991.000.99炮附0.970.780.91全蝎0.940.910.97
九香虫0.960.980.99海桐皮0.920.750.91玉竹0.970.800.97
薏仁0.970.990.99辛夷0.980.690.84生地0.980.860.97
谷虫0.980.990.99陈皮0.550.850.84芡实0.960.910.97
白芍0.971.000.99砂仁0.870.500.76
皂角0.960.890.99桑寄生0.590.410.62
), ArticleFig(id=1256183577167913665, tenantId=1146029695717560320, journalId=1255847919539208197, articleId=1256183371516993788, language=CN, label=表4, caption=

YOLOv10n-GB算法测试集各类别结果

, figureFileSmall=null, figureFileBig=null, tableContent=
类别PRmAP50类别PRmAP50类别PRmAP50
槟榔1.001.001.00钩藤0.980.950.96蝼蛄0.960.940.99
无患子0.991.000.99土元0.900.960.96虻虫0.960.950.99
蟑螂0.980.980.99枳壳0.920.840.95桂皮0.960.960.99
地龙0.921.000.99黄柏0.870.890.95蟋蟀0.981.000.99
僵蝉0.970.970.99枸杞子0.970.880.95草果0.990.800.97
粪金龟0.991.000.99炮附0.970.780.91全蝎0.940.910.97
九香虫0.960.980.99海桐皮0.920.750.91玉竹0.970.800.97
薏仁0.970.990.99辛夷0.980.690.84生地0.980.860.97
谷虫0.980.990.99陈皮0.550.850.84芡实0.960.910.97
白芍0.971.000.99砂仁0.870.500.76
皂角0.960.890.99桑寄生0.590.410.62
)], attaches=null, journal=Journal(id=1255846881855815682, delFlag=0, nameCn=中华中医药学刊, nameEn=Chinese Archives of Traditional Chinese Medicine, nameHistory1=null, nameHistory2=null, issn=1673-7717, eissn=null, cn=21-1546/R, coden=null, periodic=0, language=CN, oaType=null, ccby=null, superviseOffice=null, ownerOffice=null, pubOffice=null, editorOffice=null, officeType=null, aims=null, clcCode=null, officeProv=null, officeCity=null, officeAddr=null, officeZip=null, officeEmail=null, officePhone=null, editDirector=null, officeDirector=null, officeDirectorPhone=null, officeStaffNum=null, officeEmpNum=null, coverPicUrl=v4T+dd+NLAnioIpGAbkcEQ==, journalPrice=null, startedYear=null, abbrevIsoEn=Chinese Archives of Traditional Chinese Medicine, journalRemark=null, publicationField=null, createdTime=1777346829146, updatedTime=1777347949387, createdBy=18614031015, updatedBy=13701087609, firstLetterCn=C, firstLetterEn=C, subjectCode=Life Sciences, subjectName=null, subjectCodeEn=Life Sciences, subjectNameEn=null, picCn=v4T+dd+NLAnioIpGAbkcEQ==, picEn=RAy5vuC40hwJB+Y0pEARug==, jcr=null, cjcr=null, exts=[JournalExt(id=1255851580587368789, language=CN, name=中华中医药学刊, nameHistory1=null, nameHistory2=null, managedBy=, sponsoredBy=, publishedBy=, editorOffice=, officeProv=null, officeCity=null, officeAddr=, officeZip=, editDirector=, officeDirector=null, officePhone=null, coverPicUrl=null, journalRemark=, submitArticleUrl=null, websiteUrl=, createdTime=1777347949406, updatedTime=1777347949406, createdBy=13701087609, updatedBy=13701087609, submissionGuidelinesUrl=, submissionAuthorUrl=https://zyhs.cbpt.cnki.net/index.aspx?t=1, submissionEditorUrl=https://zyhs.cbpt.cnki.net/index.aspx?t=3, submissionReviewUrl=https://zyhs.cbpt.cnki.net/index.aspx?t=2, submissionCeEditorUrl=, submissionAeEditorUrl=, option={"copyright":""}), JournalExt(id=1255851580641894742, language=EN, name=Chinese Archives of Traditional Chinese Medicine, nameHistory1=null, nameHistory2=null, managedBy=, sponsoredBy=, publishedBy=, editorOffice=, officeProv=null, officeCity=null, officeAddr=, officeZip=, editDirector=, officeDirector=null, officePhone=null, coverPicUrl=null, journalRemark=, submitArticleUrl=null, websiteUrl=, createdTime=1777347949419, updatedTime=1777347949419, createdBy=13701087609, updatedBy=13701087609, submissionGuidelinesUrl=, submissionAuthorUrl=https://zyhs.cbpt.cnki.net/index.aspx?t=1, submissionEditorUrl=https://zyhs.cbpt.cnki.net/index.aspx?t=3, submissionReviewUrl=https://zyhs.cbpt.cnki.net/index.aspx?t=2, submissionCeEditorUrl=, submissionAeEditorUrl=, option={"copyright":""})], databaseList=null, tenantJournalId=1255847919539208197, websiteList=[Website(id=1255847989868900719, webName=null, webTitle=null, webDomain=null, webCopyrigh=null, webIpcNo=null, seoTitle=null, seoKeywords=null, seoDescription=null, tenantJournalId=null, journalId=1255847919539208197, journalNameCn=null, journalNameEn=null, grayFlag=null, tenantId=1146029695717560320, platformId=null, journalGroupId=null, journalGroupNameCn=null, journalGroupNameEn=null, type=1, domain=https://castjournals.cast.org.cn/joweb/zhzyyxk/CN, language=CN, createTime=1777347093312, createBy=18614031015, updateTime=1777347427646, updateBy=18614031015, name=中华中医药学刊-中文, tplId=1146099689490845704, title=中华中医药学刊, delFlag=0, indexPage=/home, props=[WebsiteProps(id=1255849512594170105, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1255847989868900719, code=articleTextType, value=kx, createTime=1777347456358, updateTime=1777347456358, creator=18614031015, updator=18614031015), WebsiteProps(id=1255849512564809974, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1255847989868900719, code=banner, value=null, createTime=1777347456352, updateTime=1777347456352, creator=18614031015, updator=18614031015), WebsiteProps(id=1255849512631918844, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1255847989868900719, code=grayFlag, value=0, createTime=1777347456367, updateTime=1777347456367, creator=18614031015, updator=18614031015), WebsiteProps(id=1255849512552227061, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1255847989868900719, code=logo, value=https://castjournals.cast.org.cn/joweb/zhzyyxk/CN/file/pic?fileId=uitCA2DBwhJaQSiSV07uhw==, createTime=1777347456348, updateTime=1777347456348, creator=18614031015, updator=18614031015), WebsiteProps(id=1255849512657084670, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1255847989868900719, code=minRunFlag, value=0, createTime=1777347456373, updateTime=1777347456373, creator=18614031015, updator=18614031015), WebsiteProps(id=1255849512589975800, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1255847989868900719, code=picServerUrl, value=https://castjournals.cast.org.cn/joweb/zhzyyxk/CN/file/pic, createTime=1777347456357, updateTime=1777347456357, creator=18614031015, updator=18614031015), WebsiteProps(id=1255849512648696061, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1255847989868900719, code=silenceFlag, value=0, createTime=1777347456371, updateTime=1777347456371, creator=18614031015, updator=18614031015), WebsiteProps(id=1255849512581587191, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1255847989868900719, code=staticResourcePath, value=https://castjournals.cast.org.cn/joweb/cast_kjdb_cn_619/, createTime=1777347456355, updateTime=1777347456355, creator=18614031015, updator=18614031015), WebsiteProps(id=1255849512619335930, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1255847989868900719, code=themeColor, value=null, createTime=1777347456364, updateTime=1777347456364, creator=18614031015, updator=18614031015), WebsiteProps(id=1255849512627724539, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1255847989868900719, code=themeStyle, value=null, createTime=1777347456366, updateTime=1777347456366, creator=18614031015, updator=18614031015)]), Website(id=1255847989944398211, webName=null, webTitle=null, webDomain=null, webCopyrigh=null, webIpcNo=null, seoTitle=null, seoKeywords=null, seoDescription=null, tenantJournalId=null, journalId=1255847919539208197, journalNameCn=null, journalNameEn=null, grayFlag=null, tenantId=1146029695717560320, platformId=null, journalGroupId=null, journalGroupNameCn=null, journalGroupNameEn=null, type=1, domain=https://castjournals.cast.org.cn/joweb/zhzyyxk/EN, language=EN, createTime=1777347093330, createBy=18614031015, updateTime=1777347417860, updateBy=18614031015, name=中华中医药学刊-英文, tplId=1146101810881728533, title=Chinese Archives of Traditional Chinese Medicine, delFlag=0, indexPage=/home, props=[WebsiteProps(id=1255849538993119491, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1255847989944398211, code=articleTextType, value=kx, createTime=1777347462652, updateTime=1777347462652, creator=18614031015, updator=18614031015), WebsiteProps(id=1255849538972147968, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1255847989944398211, code=banner, value=null, createTime=1777347462647, updateTime=1777347462647, creator=18614031015, updator=18614031015), WebsiteProps(id=1255849539047645446, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1255847989944398211, code=grayFlag, value=0, createTime=1777347462665, updateTime=1777347462665, creator=18614031015, updator=18614031015), WebsiteProps(id=1255849538930204927, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1255847989944398211, code=logo, value=https://castjournals.cast.org.cn/joweb/zhzyyxk/EN/file/pic?fileId=uitCA2DBwhJaQSiSV07uhw==, createTime=1777347462637, updateTime=1777347462637, creator=18614031015, updator=18614031015), WebsiteProps(id=1255849539064422664, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1255847989944398211, code=minRunFlag, value=0, createTime=1777347462669, updateTime=1777347462669, creator=18614031015, updator=18614031015), WebsiteProps(id=1255849538988925186, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1255847989944398211, code=picServerUrl, value=https://castjournals.cast.org.cn/joweb/zhzyyxk/EN/file/pic, createTime=1777347462651, updateTime=1777347462651, creator=18614031015, updator=18614031015), WebsiteProps(id=1255849539060228359, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1255847989944398211, code=silenceFlag, value=0, createTime=1777347462668, updateTime=1777347462668, creator=18614031015, updator=18614031015), WebsiteProps(id=1255849538980536577, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1255847989944398211, code=staticResourcePath, value=https://castjournals.cast.org.cn/joweb/cast_kjdb_en_623/, createTime=1777347462649, updateTime=1777347462649, creator=18614031015, updator=18614031015), WebsiteProps(id=1255849539001508100, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1255847989944398211, code=themeColor, value=null, createTime=1777347462654, updateTime=1777347462654, creator=18614031015, updator=18614031015), WebsiteProps(id=1255849539039256837, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1255847989944398211, code=themeStyle, value=null, createTime=1777347462663, updateTime=1777347462663, creator=18614031015, updator=18614031015)])], journalTitle=中华中医药学刊, weixinUrl=null, journalUrl=https://castjournals.cast.org.cn/zhzyyxk, iacademicId=null, status=1, seqNo=null, journalTitleEn=Chinese Archives of Traditional Chinese Medicine, journalPhotoCn=v4T+dd+NLAnioIpGAbkcEQ==, journalPhotoEn=RAy5vuC40hwJB+Y0pEARug==, journalFirstLetter=C, journalRecommend=null, journalNew=null, journalCollection=null, jcrJf=null, cjcrJf=null, jcrJfStr=null, cjcrJfStr=null, submissionFirstDecision=null, sciSubjectClassification=null, casSubjectClassification=null, citeScore=null, totalCitationFrequency=null, icpCode=null, psCode=null, advertisingLicenseCode=null, copyrightInformation=null, country=null, option=, provinceCode=null, provinceName=null, collectFlag=false), detailUrlCn=https://castjournals.cast.org.cn/joweb/zhzyyxk/CN/10.13193/j.issn.1673-7717.2025.12.004, detailUrlEn=https://castjournals.cast.org.cn/joweb/zhzyyxk/EN/10.13193/j.issn.1673-7717.2025.12.004, pdfUrlCn=https://castjournals.cast.org.cn/joweb/zhzyyxk/CN/PDF/10.13193/j.issn.1673-7717.2025.12.004, pdfUrlEn=https://castjournals.cast.org.cn/joweb/zhzyyxk/EN/PDF/10.13193/j.issn.1673-7717.2025.12.004, aliStartDate=null, aliEndDate=null, collectionFlag=false, citedCount=null, citedUrl=null, reference=null)
收藏切换
面向移动端的轻量化YOLOv10中药材检测算法研究
收藏切换
PDF下载
周伟杰 1 , 庞立健 2 , 吕晓东 1 , 程彦杰 1 , 谢晓雯 1 , 邱巾格 1
中华中医药学刊 | 数字中医中药 2025,43(12): 21-25
收起
收藏切换
中华中医药学刊 | 数字中医中药 2025, 43(12): 21-25
面向移动端的轻量化YOLOv10中药材检测算法研究
全屏
周伟杰1, 庞立健2, 吕晓东1, 程彦杰1, 谢晓雯1, 邱巾格1
作者信息
  • 1.辽宁中医药大学,辽宁 沈阳 110847
  • 2.辽宁中医药大学附属医院,辽宁 沈阳 110032
  • 周伟杰(1994-),男,辽宁大连人,助教,硕士,研究方向:中医人工智能、机器视觉、中医大数据分析。

通讯作者:

吕晓东(1966-),女,辽宁沈阳人,教授,博士研究生导师,博士,研究方向:中医肺病与络病基础与临床。
Research on Mobile Oriented Lightweight YOLOv10 Algorithm for Chinese Herbal Medicine Detection
Weijie ZHOU1, Lijian PANG2, Xiaodong LYU1, Yanjie CHENG1, Xiaowen XIE1, Jinge QIU1
Affiliations
  • 1.Liaoning University of Traditional Chinese Medicine,Shenyang 110847,Liaoning,China
  • 2.Liaoning University of Traditional Chinese Medicine Affiliated Hospital,Shenyang 110032,Liaoning,China
出版时间: 2025-12-10 doi: 10.13193/j.issn.1673-7717.2025.12.004
文章导航
收藏切换

利用轻量化的深度学习YOLO(You Only Look Once,YOLO)v10目标检测算法,构建了面向移动端的高效中药材检测算法,为中药材种植生产、质量评价、教学科普以及自动化抓药等工作场景中的中药材检测提供了智能、高效的技术方案。建立包含31种常用中药材目标检测数据集,共6900张图像。为提升模型的多尺度检测能力和检测效率,引入幻影卷积(Ghost Convolution,GhostConv)轻量化卷积与加权双向特征金字塔模块(Bidirectional Feature Pyramid Network,BiFPN),提出了一种改进的轻量化中药材检测模型YOLOv10n-GB,并利用算法对数据集中图像样本进行训练、测试、分析、移动端部署测试。改进后模型的参数量2.27M(Million,M)、计算量6.4G(Giga,G)、平均精度mAP50(mean Average Precision 50,mAP50)为0.947,与YOLOv10n相比参数量降低1.2%、计算量降低4.5%,mAP50提升了1.8%,在移动端的模型检测平均帧率8.1帧每秒(Frames Per Second,FPS)。与其他轻量化算法YOLOv5n、YOLOv8n相比,YOLOv10n-GB具有最低的浮点计算量和更高的检测精度。YOLOv10n-GB算法完成了面向移动端的中药材的高效检测任务,为建立便携、实时的中药材检测方案开辟了新思路。

深度学习  /  目标检测  /  YOLOv10  /  BiFPN  /  GhostConv  /  中药材检测

In this research,the lightweight deep learning YOLOv10 object detection algorithm is harnessed to develop an efficient algorithm for detecting Chinese herbal medicines on mobile devices,furnishing an intelligent and high-performance technical solution for the detection of Chinese herbal medicines in scenarios such as cultivation production,quality assessment,educational popularization and automated dispensing.A dataset for target detection of 31 commonly used Chinese herbal medicines is constructed,comprising a total of 6900 images.To enhance the multi-scale detection capability and detection efficiency of the model,the Ghost Convolution(Ghost Convolution,Ghost Conv)lightweight convolution and the weighted bidirectional feature pyramid module(Bidirectional Feature Pyramid Network,BiFPN)are incorporated,and an improved lightweight model for Chinese herbal medicine detection,namely YOLOv10n-GB,is proposed.The algorithm is then employed to conduct training,testing,analysis and mobile terminal deployment tests on the image samples within the dataset.The improved model exhibits2.27 M(Million)parameters,6.4 G(Giga)computational complexity,and an m AP50 value of0.947.In comparison with YOLOv10n,the parameter count is reduced by 1.2%,the computational load is decreased by 4.5%,and the mean average precision 50(mAP50)is augmented by 1.8%.The average frame rate of the model detection on the mobile terminal reaches 8.1Frames Per Second(FPS).When contrasted with other lightweight algorithms such as YOLOv5n and YOLOv8n,YOLOv10n-GB demonstrates the lowest floating-point computational cost and higher detection accuracy.The YOLOv10n-GB algorithm has accomplished the task of efficient detection of Chinese herbal medicines on mobile terminals,paving the way for novel ideas in establishing a portable and real-time detection scheme for Chinese herbal medicines.

deep learning  /  object detection  /  YOLOv10  /  BiFPN  /  GhostConv  /  Chinese herbal medicine detection
周伟杰, 庞立健, 吕晓东, 程彦杰, 谢晓雯, 邱巾格. 面向移动端的轻量化YOLOv10中药材检测算法研究. 中华中医药学刊, 2025 , 43 (12) : 21 -25 . DOI: 10.13193/j.issn.1673-7717.2025.12.004
Weijie ZHOU, Lijian PANG, Xiaodong LYU, Yanjie CHENG, Xiaowen XIE, Jinge QIU. Research on Mobile Oriented Lightweight YOLOv10 Algorithm for Chinese Herbal Medicine Detection[J]. Chinese Archives of Traditional Chinese Medicine, 2025 , 43 (12) : 21 -25 . DOI: 10.13193/j.issn.1673-7717.2025.12.004
中药材作为中医的核心支柱,可以分为植物、动物、矿物、贝壳等多种形态[1]。传统手工鉴别方法耗费人力,且具有人为主观因素,为了提升鉴别效率,目前计算机视觉、图像处理等技术被逐步运用于中药鉴定中[2]
目前深度学习技术应用于中药鉴别领域已有一定进展[3]。张琦运用Xception网络,在包含13 088张图像的数据集上实现了97.26%的准确率[4]。李玥辰则在AlexNet基础上创新性地融入了SE注意力机制与Mish激活函数,针对50种常见中药材的1090张图像,实现了测试集top-1高达98.3%的准确率[5]。周丽媛通过多特征融合结合注意力机制,对20种中药饮片及复杂背景下的32种藏药材切片图像进行识别,分别取得了98.00%和91.68%的精度[6]。谷瑞在Res2Net网络中引入通道注意力机制模块,在16种常见中药饮片自有数据集上的检测准确率为94.74%[7]。高爽使用YOLOv3算法在152种中药饮片图像数据集上的平均识别精度为96.2%[8]。郭丛则构建了针对108种饮片的YOLOv4目标检测模型,平均识别精度mAP达到85.3%[9]。华畅构建了自有63种6853幅图像的中药饮片数据集,并使用YOLOv8目标检测算法在测试数据集上的准确率为96.5%[10]
上述深度学习算法在中药材识别领域表现优异,但多数算法的计算量和参数量较大,制约其在资源受限的移动设备上的部署能力,从而影响了移动端中药材识别的应用。为此,本研究采用轻量化目标检测算法,旨在应用于移动端的中药材检测任务。本研究旨在为用户提供一个便捷、高效的移动端中药材检测方案,显著提升中药材的鉴别效率。
本研究所采用的药材图像样本均采集于实验室环境中,通过高分辨率手机拍摄而成,以确保图像质量的清晰与一致性,为后续实验分析提供了可靠的数据基础。数据集中包含动物与植物共计31类。具体类别如下:蝼蛄、砂仁、槟榔、玉竹、桂皮、辛夷、土元、蟑螂、薏仁、黄柏、蟋蟀、粪金龟、地龙、芡实、枳壳、虻虫、皂角、草果、谷虫、僵蝉、白芍、桑寄生、炮附、全蝎、无患、九香虫、钩藤、枸杞子、海桐皮、生地、陈皮。数据集中的部分目标如插页Ⅱ图1a所示。本研究为每种中药材采集了200~300幅图像,构建了一个包含6900幅图像的数据集,涵盖了63 666个标注目标。这些图像充分考虑了中药材的多样性,确保了数据集的广泛性和代表性,各类中药材目标在数据集中的数量分布情况如插页Ⅱ图1b所示。数据集中的目标尺度多样,目标框的尺度主要集中在0.2以下,既包含枸杞、芡实、薏苡仁等小目标,也包括皂角、全蝎等较大目标。目标框尺度归一化分布详见插页Ⅱ图1c
数据集制备流程从药材选择开始,首先通过手机拍摄图像,然后使用LabelImg手工标注中药材目标。为提升标注效率,本研究引入yolov8-resnet-101[11-12]目标检测算法完成数据预标注任务,并最终由人工对模型标注的数据进行微调,以保证数据标注的质量。数据集规划阶段,最后将数据按7∶1.5∶1.5的比值分为训练集、验证集和测试集。
目标检测是计算机视觉领域中的一个重要方向[13]。基于深度学习的目标检测算法是给出图像中特定目标类别和具体坐标的算法。目前出现的算法大致分为两类:一类是基于候选区域的二阶段目标检测算法:另一类是基于回归的一阶段目标检测算法[14],二者算法的特点见表1。最具代表性的二阶段算法是Faster R-CNN,它检测精度高、速度慢、参数量大,很难部署在移动端。一阶段算法舍弃二阶段的目标候选框策略,直接通过回归方法定位到目标的边界框的坐标信息,其中YOLO系列算法最具代表性。YOLO通过不断优化模型,从YOLOv1注重检测速度到现在YOLOv10模型整体架构、平衡速度、精度和模型大小,每一次的版本迭代都在解决以往YOLO系列的不足[15],并且YOLOv10对轻量化部署做出了诸多优化方案:如部分自注意力机制(Partial Self-Attention,PSA)、去除非极大值抑制(Non-Maximum Suppression,NMS)后处理等。因此,本研究采用YOLOv10算法进行移动端中药材的检测任务。
清华大学提出的YOLOv10目标检测算法涵盖n至x的多种方案,其中YOLOv10n是专为移动端设计的轻量化模型。YOLOv10n的结构包含以下四个部分:数据输入、Backbone主干网络、Neck多尺度特征融合、双重标签检测头[25],网络结构如插页Ⅲ图2所示。数据输入结构提供数据增强的接口,如:训练图像的多尺度缩放、平移、翻转、随机块状马赛克等。Backbone主干网络:YOLOv10n继承了YOLOv8的带有两次卷积和融合操作的跨阶段部分网络(Cross Stage Partial Network with 2 convolutions and a Fusion,C2F)与空间金字塔池化快速模块(Spatial Pyramid Pooling Fast,SPFF)网络结构,同时也添加了独有的步幅卷积下采样(Strided Convolution Downsampling,SCDown)、PSA、带有两次卷积、融合操作和上下文信息瓶颈的跨阶段部分网络(Cross Stage Partial Network with 2 convolutions,a Fusion,and Contextual Information Bottleneck,C2fCIB)网络结构。Neck结构采用图像金字塔的方法实现多尺度特征融合,进而提升小目标的检测精度。双重标签检测头结构利用One-to-one Head与Oneto-many Head轻量级检测头进行联合优化,使Backbone与Neck结构提取到的特征受到更多的信号监督,在模型在前向推理阶段,One-to-many Head被舍弃,仅使用One-to-one Head输出目标数值,不需要NMS后处理机制,从而起到加速模型推理的作用。损失函数采用解耦头处理分类、回归与置信度,以提升检测效率,完全交并比(Complete Intersection over Union,CIoU)损失函数综合考虑重叠、中心距离与纵横比,增强了目标框的定位精度。
YOLOv10n引入了诸多轻量化设计:SCDown是一种空间和通道解耦的降维方法,首先逐点卷积变更通道数,然后通过深度卷积进行空间特征降维,这种方法在降低计算成本的同时,最大限度地保留了信息。相比传统的下采样方法,SCDown能够更有效地提取特征,同时减少计算量。PSA是一种高效的局部自注意力模块,它将卷积后的特征一部分输入到由多头自注意力模块(Multi-Head Self-Attention,MHSA)和前馈网络(Feed-Forward Network,FFN)组成的模块中,然后将两部分连接起来并通过卷积进行融合,这种结构能够增强全局体征提取的能力,同时降低算法的计算复杂度。C2fCIB是在C2 f模块的基础上,用紧凑的倒置块(Compact Inverted Block,CIB)替换了原本的Bottleneck结构。CIB将Bottleneck中的标准卷积用深度卷积加逐点卷积进行替换,从而降低了计算量并增大了感受野。
考虑到本研究中的目标检测算法需部署在移动设备上,模型的参数量和计算量都需要较压缩。Yolov10n算法已经做了大量的轻量化设计,例如C2F模块、SPFF模块,SCDown模块来提升模型的检测效率,但YOLOv10n网络中的第2、4、18层使用普通卷积完成特征提取和降维,这些卷积位置靠前,占用了较多计算资源,并且产生诸多的冗余特征图。为提升模型特征提取的效率,本研究引入华为诺亚方舟实验室提出的轻量化GhostConv结构[26],其特征提取结构如插页Ⅱ图3所示:首先,GhostConv通过窄通道普通提取特征,其次,采用计算量较低的深度可分离卷积提取线性特征,最后,将窄通道特征图进行恒等映射与线性特征进行合并,这种廉价计算能够得到与普通卷积类似的特征图,从而提升算法在中药材检测任务的效率。
普通卷积的计算量FLOPs如公式(1)所示:
GhostConv的计算量FLOPs'如公式(2)所示:
普通卷积的参数量Prams计算如公式(3)所示:
GhostConv的计算量Prams'如公式(4)所示:
普通卷积与GhostConv的计算量压缩比计算如公式(5)所示:
普通卷积与GhostConv的参数量压缩比计算如公式(6)所示:
上述公式中:c代表中药材特征图的输入的通道数,ho、wo、no代表中药材特征图的输出的高度、宽度、通道数,k代表普通卷积核的尺寸,d代表第二步提取线性中药材特征的卷积核尺寸,s(s<n)代表第二步中药材线性特征图输出的维度。
通过对两个公式(5)与(6)可知Ghostconv与普通卷积的计算量与参数量压缩比无限趋近s,可以证明采用Ghostconv卷积可以减少模型的参数量和计算量。
中药材目标检测数据集的种类繁多,尺寸和形状各异,且在不同尺度下展现出不同的纹理特征,这给中药材的检测工作带来了挑战,导致模型在精确识别中药材时面临困难。为了克服中药材尺寸差异所带来的精度问题,本研究对YOLOv10n的Neck多尺度特征融合进行了优化,引入了高效双向跨尺度连接和加权特征融合思想,将Neck部分中concat层改进为谷歌研究院提出的基于双向跨尺度连接和加权特征融合BiFPN模块[27],网络结构如插页Ⅱ图4a所示。BiFPN模块通过优化特征融合路径,去除了只有单一输入的节点以降低特征冗余,并采用多次重复的双向路径来实现更深层次的特征融合和信息传递,提升多尺度和复杂外观中药材的检测效果;此外,BiFPN模块还引入了基于特征重要性的加权融合策略,以提升特征融合效果。同时在同层特征之间增加了残差连接,这样做可以在不增加额外计算负担的前提下,尽可能地融合更多特征,减少细节特征信息的丢失。为了使模型更适合在移动设备上部署,本研究选择仅保留P3、P4、P5 3个尺度的特征,这种策略能够在保持特征融合效果的同时,进一步简化网络结构,从而使模型更加精简和轻量化,Concat_BiFPN的网络结构如插页Ⅱ图4b所示。
模型在特征图融合阶段,不同分辨率的中药材特征图针对不同尺度的目标贡献度不同,本研究所使用的中药材数据集包含了大量的多尺度目标,本研究通过Concat_BiFPN模块可以快速融合不同尺寸特征图,提取中药材特征图的深层信息,解决因目标尺度引起的精确度与召回率降低的问题。Concat_BiFPN模块的加权特征融合公式如下:
公式(7)中字符含义如下:Output为输出中药材特征,wi为第i层输入特征的权重,Inputi为第i层输入的中药材特征,W 为全部特征总和,ε用于约束数值振荡的极小值学习率,默认值为0.0001。
实验硬件:CPU为Intel(R)Xeon(R)Platinum 8358P CPU@ 2.60GHz,GPU为NVIDIA RTX 3090,显存24GB,内存DDR5 90GB。实验软件:Ubuntu 18.04操作系统,Python版本为3.8,Pytorch版本为1.9.0,CUDA的版本为11.0。移动设备为华为P70,CPU为麒麟9000S处理器,八核心架构,运行内存12G。
采用的评估指标有:精确度(Precision,P)、召回率(Recall,R)、平均精度mAP、权重参数量(单位MB)、检测平均帧率(FPS)与计算量(G单位),评价指标的计算方法如公式(8)、(9)、(10)、(11)所示:
上述公式中,TP表示真正例,TN表示真负例,FP表示假正例,FN表示假负例。P为精确度,表示正确预测的正例占所有预测为正例的比例。R为召回率,表示正确预测的正例占所有实际正例的比例。AP值是根据P-R曲线计算的面积,AP指单个类别平均精度,mAP是所有类别的平均精度,mAP50是当预测边界框与真实边界框的重叠程度大于等于50%的平均精度,n表示总类别数。参数量衡量算法的总能参数的量,以MB为单位。计算量衡量算法的计算复杂度以及处理器每秒能执行的操作次数,以G为单位。平均帧率(FPS)用于衡量模型每秒推理的图像帧数,反映模型运行效率和实时性能。
为了验证针对YOLOv10n在中药材目标检测数据集上的改进策略的有效性,对每个策略都进行消融实验,不同改进方法对算法检测性能如表2所示。
表2消融实验结果显示:基础模型中引入GhostConv轻量化卷积后组成YOLOv10n-G算法,算法参数量降低1.2%、计算量降低4.5%,mAP50提升0.5%。基础模型中引入加权双向特征金字塔模块组成YOLOv10n-B算法,算法参数量与计算量基本不变,mAP50提升了1.6%,表明Concat_bifpn模块的引入显著提升了模型对多尺度中药材的特征提取能力。引入两种优化方案叠加组成YOLOv10n-GB算法,改进后算法和YOLOv10n-B的参数量、计算量一致,并且mAP50提升了1.8%,显著提升中药材检测效果。通过分析后得知:本研究提出的轻量化模型YOLOv10n-GB比YOLOv10n具备更好的检测效率。
为验证本研究改进YOLOv10n算法的高效性,本研究还将其与其他轻量化目标检测算法进行了对比实验,结果如表3所示。
本研究设计涉及的算法包括:YOLOv5n、YOLOv8n、YOLOv10n、改进算法YOLOv10n-GB。从表3数据可以看出,YOLOv10n-GB在精确度上达到了0.932,召回率为0.884,mAP50为0.947,除了召回率均优于YOLOv10n、YOLOv8n、YOLOv5n,这表明本研究的YOLOv10n-GB算法在保持高检测帧率的同时,提高了精确度。在参数量方面,YOLOv10n-GB为2.27MB,低于YOLOv10n的2.29MB、YOLOv8n的3.2MB、YOLOv5n的2.7MB,表明YOLOv10n-GB算法在参数量最少。在计算复杂度方面,YOLOv10n-GB的FLOPs为6.4G,低于其他算法的6.7G、8.3G、8.7G,显示出YOLOv10n-GB算法在计算效率上有所提升。YOLOv10n-GB的平均帧率为769FPS,与YOLOv5n和YOLOv8n相比提升明显,YOLOv10n-GB在轻量化目标检测算法中表现出了优异的性能,不仅在精确度和召回率上有所提升,而且在计算效率上也有所优化,这为移动端实时中药材检测任务提供了一个有效的解决方案。
在本研究中,YOLOv10n-GB算法在中药材目标检测任务中表现出色,具体检测结果详见表4。分析结果表明,高达80%的中药材类别在mAP50上超过了0.95,并且87%的类别mAP50超过了0.9,这证明了YOLOv10n-GB在测试集上具有极高的检测精度。特别是,槟榔、无患子、地龙、白芍和粪金龟等类别的召回率达到了1,这表明算法能够完全检测出这些类别的所有目标实例。此外,61.2%的中药材类别在精确度和召回率两个指标上均达到了0.9以上,这说明算法在维持高精确度的同时,也能有效减少漏检情况。然而,仍有部分类别的精确度和召回率未能达到0.9,例如陈皮,该类别目标尺度的多样性和外观形状的多变性可能是导致检测精度较低的主要原因。尽管YOLOv10n-GB算法在大多数中药材类别上都取得了良好的性能,但仍存在提升空间。未来的工作将聚焦于进一步提升算法对特定类别的检测性能,如针对陈皮与桑寄生等,可以通过增加多样化的数据来优化算法性能。
本研究开发了面向移动终端的中药材检测系统,为提升移动端的检测效率,在部署阶段调用pytorch的optimize_for_mobile接口优化模型。系统实现拍照检测,照片检测、视频检测、统计药材检测数量等功能。为验证优化后YOLOv10n-GB的推理速度,系统随机抽取测试集中20幅图像统计推理时长,计算得出检测平均帧率为8.1FPS,可以满足系统实时性的要求,安卓移动终端APP检测效果如插页Ⅱ图5所示。
为提升移动端多尺度中药材检测的效率,本研究提出了一种基于YOLOv10n的改进算法YOLOv10n-GB。该算法基于YOLOv10n网络结构,引入加权双向特征金字塔网络Concat_BiFPN网络模块,将多尺度的中药材特征高效融合,提升了算法检测精度;将普通卷积模块替换成GhostConv以降低模型参数和计算量,提高移动端的中药材检测效率;创建了一个包含31种中药材检测信息的数据集,这为算法在工程实践中的应用提供了坚实的基础。在中药材数据集上进行实验,结果表明:YOLOv10n-GB比原YOLOv10n算法的mAP提高了1.8%,参数量降低1.2%、计算量降低4.5%。与其他主流轻量化算法进行对比,YOLOv10n-GB的检测精度更高且速度更快。YOLOv10n-GB算法在安卓移动端的检测平均帧率为8.1FPS,可以实现移动端实时检测。基于YOLOv10n-GB中药材检测算法在移动端表现出卓越的性能,有助于推动中药材检测任务向便携智能化方向的快速发展。
  • 国家自然科学基金面上项目(82274440)
  • 国家中医药管理局中医络病重点学科建设项目(T0302)
  • 辽宁省中医药创新团队项目(LNZYYCXTD-CCCX-001)
参考文献 引证文献
排序方式:
[1]
冒巍巍,范莉,王方华. 我国中药产业国际竞争力分析[J]. 中草药2003(4):4-7.
[2]
徐雅静,俞捷,余远盼,. 人工智能在中药材及饮片鉴别领域的应用[J]. 中华中医药学刊202240(8):47-50.
[3]
谭代庆,肖志鸿,吴浩忠,. 融合改进残差网络和Transformer的易混淆中药饮片识别研究[J]. 中国数字医学202318(6):42-50.
[4]
张琦,区锦锋,周华英. 基于Xception与迁移学习的中药饮片图像识别研究[J]. 现代电子技术202447(3):29-33.
[5]
李玥辰,赵晓,王若男,. 基于改进卷积神经网络的中药饮片图像识别[J]. 科学技术与工程202424(9):3596-3604.
[6]
周丽媛. 基于深度学习的草本藏药材识别方法研究[D]. 拉萨:西藏大学,2023.
[7]
谷瑞,宋翠玲,李元昊. 改进Res2Net和注意力的中药饮片识别模型[J]. 国外电子测量技术202443(9):130-140.
[8]
高爽,周志强,黄显章,. 基于YOLOv3算法的中药饮片检测研究[J]. 中药材202346(1):57-61.
[9]
郭丛,田钰嘉,李杨,. 基于YOLOv4算法的中药饮片识别[J]. 中国实验方剂学杂志202329(14):133-140.
[10]
华畅,郑豪. 面向中药饮片识别的MSDA-YOLOv8检测算法[J]. 时珍国医国药202435(12):2898-2904.
[11]
VARGHESE R,SAMBATH M. YOLOv8:A Novel Object Detection Algorithm with Enhanced Performance and Robustness[C]//2024 International Conference on Advances in Data Engineering and Intelligent Computing Systems(ADICS),2024.
[12]
HE K,ZHANG X,REN S,et al. Deep residual learning for image recognition[J]. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition(CVPR)2016:770-778.
[13]
郭庆梅,刘宁波,王中训,. 基于深度学习的目标检测算法综述[J]. 探测与控制学报202345(6):10-20,26.
[14]
董文轩,梁宏涛,刘国柱,. 深度卷积应用于目标检测算法综述[J]. 计算机科学与探索202216(5):1025-1042.
[15]
黄毅,周纯,刘欣军,. 基于YOLOv10的无人机复杂背景下多尺度检测模型[J]. 光通信研究202410,30(1):1-8.
[16]
GIRSHICK R. Fast R-CNN[C]//International Conference on Computer Vision. IEEE Computer Society,2015:1440-1448.
[17]
REN S,HE K,GIRSHICK R,et al. Faster R-CNN:towards realtime object detection with region proposal networks[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence201739(6):1137-1149.
[18]
HE K M,GKIOXARIG,PIOTR D,et al. MaskR-CNN[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition(CVPR),2017:2961-2969.
[19]
PANG J,CHEN K,SHI J,et al. Libra R-CNN:towards balanced learning for object detection[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition,2019:821-830.
[20]
LU X,LIB,YU EY,et al. GridR-CNN[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition(CVPR),2019:7363-7372.
[21]
LIUW,ANGUELOV D,ERHAN D,et al. SSD:Single Shot Multi-Box Detector[C]//Proceedings of the Computer Vision(CVPR)-ECCV 2016. Springer International Publishing,2016:21-37.
[22]
WANG C Y,BOCHKOVSKIY A,LIAO H Y M. YOLOv7:Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors[C]//IEEE/CVF Conference on Computer Vision and Pattern Recognition(CVPR),Vancouver,Canada. IEEE,2023:7464-7475.
[23]
WANG C Y,YEH IH,LIAO H Y M. Yolov9:Learning what you want to learn using programmable gradient information[C]//IEEE Conference on Computer Vision and Pattern Recognition(CVPR). SeattleWA,USA:IEEE,2024.
[24]
WANG A,CHEN H,LIU L,et al. Yolov10:Real-time end-toend object detection[C]//IEEE Conference on Computer Vision and Pattern Recognition(CVPR). SeattleWA,USA:IEEE,2025.
[25]
高立鹏,周孟然,胡锋,. 基于REIW-YOLOv10n的井下安全帽小目标检测算法[J]. 煤炭科学技术2024(1):1-13.
[26]
K HAN,Y WANG,Q TIAN,et al. GhostNet:More Features From Cheap Operations[C]//2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition(CVPR),Seattle,WA,USA,2020:1577-1586.
[27]
TAN M M,PANG R,LE Q V. EfficientDet:Scalable and Efficient Object Detection[C]//2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition(CVPR),2020:10778-10787.
2025年第43卷第12期
PDF下载
58
25
引用本文
BibTeX
文章信息
doi: 10.13193/j.issn.1673-7717.2025.12.004
  • 首发时间:2026-04-29
  • 出版时间:2025-12-10
补充材料
相关文章
文章信息
作者
出版历史
基金
国家自然科学基金面上项目(82274440)
国家中医药管理局中医络病重点学科建设项目(T0302)
辽宁省中医药创新团队项目(LNZYYCXTD-CCCX-001)
作者信息
    1.辽宁中医药大学,辽宁 沈阳 110847
    2.辽宁中医药大学附属医院,辽宁 沈阳 110032

通讯作者:

吕晓东(1966-),女,辽宁沈阳人,教授,博士研究生导师,博士,研究方向:中医肺病与络病基础与临床。
参考文献
分享链接
https://castjournals.cast.org.cn/joweb/zhzyyxk/CN/10.13193/j.issn.1673-7717.2025.12.004
分享至
全文二维码

扫描看全文

引用本文
BibTeX
本文的引用情况
2种不同金属材料的力学参数

Family
属数
Number of
genus
种数
Number of
species
占总种数比例
Percentage of
total species (%)

Genus
种数
Number of
species
占总种数比例
Percentage of total
species (%)
鹅膏菌科Amanitaceae 2 11 5.26 鹅膏菌属 Amanita 10 4.78
小菇科 Mycenaceae 2 12 5.74 丝盖伞属 Inocybe 5 2.39
多孔菌科 Polyporaceae 8 14 6.70 蜡蘑属 Laccaria 5 2.39
红菇科 Russulaceae 3 23 11.00 小皮伞属 Marasmius 6 2.87
小菇属 Mycena 11 5.26
光柄菇属 Pluteus 5 2.39
红菇属 Russula 17 8.13
栓菌属 Trametes 5 2.39
关闭全屏