基于PP-PicoDet-XS的改进铝型材表面缺陷检测算法

TP391.4; 铝型材在生产加工过程中会产生特征不明显和尺度大小不一等多类型的表面缺陷,针对现有人工抽检方法准确率低、实时性差、主观性强等问题,提出一种基于PP-PicoDet-XS的改进铝型材表面缺陷检测算法.改进的算法在主干网络中嵌入无参注意力SimAM,增强对深层有效特征的提取能力;使用SIoU(Scylla intersection over union)损失函数对训练过程进行优化,提高预测框的定位能力;采用量化蒸馏策略对模型进行压缩,提高推理速度.结果表明,改进的算法平均精度均值在交并比(intersection over union,IoU)阈值为0.5时达到了98.93%,在I...

Full description

Saved in:
Bibliographic Details
Published in东北大学学报（自然科学版） Vol. 45; no. 11; pp. 1557 - 1564
Main Authors 马淑华, 李立振, 秦汉民, 沙晓鹏
Format Journal Article
LanguageChinese
Published 东北大学秦皇岛分校 控制工程学院,河北 秦皇岛 066004 15.11.2024
Subjects
损失函数
SimAM
aluminum profiles
loss function
distillation
量化
蒸馏
defects detection
铝型材
缺陷检测
quantization
Online AccessGet full text
ISSN1005-3026
DOI10.12068/j.issn.1005-3026.2024.11.005

Cover

Abstract TP391.4; 铝型材在生产加工过程中会产生特征不明显和尺度大小不一等多类型的表面缺陷,针对现有人工抽检方法准确率低、实时性差、主观性强等问题,提出一种基于PP-PicoDet-XS的改进铝型材表面缺陷检测算法.改进的算法在主干网络中嵌入无参注意力SimAM,增强对深层有效特征的提取能力;使用SIoU(Scylla intersection over union)损失函数对训练过程进行优化,提高预测框的定位能力;采用量化蒸馏策略对模型进行压缩,提高推理速度.结果表明,改进的算法平均精度均值在交并比(intersection over union,IoU)阈值为0.5时达到了98.93%,在IoU阈值0.5～0.95范围内达到了57.60%,较未压缩的原始模型分别提高了1.73%和4.13%.将该算法部署到骁龙865移动端平台上进行推理,推理速度可达116.82帧/s,较未压缩的原始模型提高了47帧/s.
AbstractList TP391.4; 铝型材在生产加工过程中会产生特征不明显和尺度大小不一等多类型的表面缺陷,针对现有人工抽检方法准确率低、实时性差、主观性强等问题,提出一种基于PP-PicoDet-XS的改进铝型材表面缺陷检测算法.改进的算法在主干网络中嵌入无参注意力SimAM,增强对深层有效特征的提取能力;使用SIoU(Scylla intersection over union)损失函数对训练过程进行优化,提高预测框的定位能力;采用量化蒸馏策略对模型进行压缩,提高推理速度.结果表明,改进的算法平均精度均值在交并比(intersection over union,IoU)阈值为0.5时达到了98.93%,在IoU阈值0.5～0.95范围内达到了57.60%,较未压缩的原始模型分别提高了1.73%和4.13%.将该算法部署到骁龙865移动端平台上进行推理,推理速度可达116.82帧/s,较未压缩的原始模型提高了47帧/s.
Abstract_FL During the production and processing of aluminum profiles,multiple types of surface defects such as unclear features and varying scales may generate.In response to the problems of low accuracy,poor real-time performance,and strong subjectivity in existing manual sampling method,an improved surface defects detection algorithm is proposed for aluminum profiles based on PP-PicoDet-XS.The SimAM attention was embedded in the backbone to enhance the ability of extracting deep effective features.The SIoU(Scylla intersection over union)loss function is used to optimize the training process to improve the positioning ability of the prediction boxes.The quantization and distillation were used to compress the model to improve the inference speed.The results show that the improved algorithm achieves a mean average precision of 98.93%at intersection over union(IoU)threshold of 0.5,and 57.60%across IoU thresholds ranging from 0.5 to 0.95,which is 1.73%and 4.13%higher than the uncompressed original model.Deploying this algorithm on the Snapdragon 865 mobile platform for inference,the inference speed can reach 116.82 frames per second,which is 47 frames per second higher than the uncompressed original model.
Author 马淑华
李立振
沙晓鹏
秦汉民
AuthorAffiliation 东北大学秦皇岛分校 控制工程学院,河北 秦皇岛 066004
AuthorAffiliation_xml – name: 东北大学秦皇岛分校 控制工程学院,河北 秦皇岛 066004
Author_FL QIN Han-min
MA Shu-hua
LI Li-zhen
SHA Xiao-peng
Author_FL_xml – sequence: 1
  fullname: MA Shu-hua
– sequence: 2
  fullname: LI Li-zhen
– sequence: 3
  fullname: QIN Han-min
– sequence: 4
  fullname: SHA Xiao-peng
Author_xml – sequence: 1
  fullname: 马淑华
– sequence: 2
  fullname: 李立振
– sequence: 3
  fullname: 秦汉民
– sequence: 4
  fullname: 沙晓鹏
BookMark eNrjYmDJy89LZWBQNTTQMzQyMLPQz9LLLC7O0zM0MDDVNTYwMtMzMjAy0TM01AMKsDBwwsU5GHiLizOTDAwMLE3MTY0sORlsn87f9WRXX0CAbkBmcr5LaoluRPDzWS3Ppux8sX_2y8lzn87rfjZ3wouFK17OXfR8z66XM7c_W9zwbGv383XTn22eysPAmpaYU5zKC6W5GULdXEOcPXR9_N09nR19dIsNDYxMdY2MLBNTzQyMklJMEg1SLS2TLcwNLIxMTEySDAwtDdMMTVPNLE2NgNAg2dg0zczUMjnNKCk51dwoxSTVItkkyZibQR1ibnliXlpiXnp8Vn5pUR7QxviUpJSKiiSQdw1BvjQGAM8QXuk
ClassificationCodes TP391.4
ContentType Journal Article
Copyright Copyright © Wanfang Data Co. Ltd. All Rights Reserved.
Copyright_xml – notice: Copyright © Wanfang Data Co. Ltd. All Rights Reserved.
DBID 2B.
4A8
92I
93N
PSX
TCJ
DOI 10.12068/j.issn.1005-3026.2024.11.005
DatabaseName Wanfang Data Journals - Hong Kong
WANFANG Data Centre
Wanfang Data Journals
万方数据期刊 - 香港版
China Online Journals (COJ)
China Online Journals (COJ)
DatabaseTitleList
DeliveryMethod fulltext_linktorsrc
Discipline Engineering
DocumentTitle_FL Improved Surface Defects Detection Algorithm for Aluminum Profiles Based on PP-PicoDet-XS
EndPage 1564
ExternalDocumentID dbdxxb202411005
GrantInformation_xml – fundername: 河北省自然科学基金资助项目
  funderid: (F2021501021)
GroupedDBID -03
2B.
4A8
5XA
5XD
92E
92I
93N
ABDBF
ABJNI
ACGFS
ACUHS
ALMA_UNASSIGNED_HOLDINGS
CCEZO
CEKLB
CW9
EAD
EAP
EAS
EOJEC
ESX
OBODZ
PSX
TCJ
TGP
U1G
U5M
ID FETCH-LOGICAL-s1025-229ae602bd4a0e99c87082444b0191f15e69525250c35f659cf2bce72d4e8c4b3
ISSN 1005-3026
IngestDate Thu May 29 03:59:15 EDT 2025
IsPeerReviewed false
IsScholarly true
Issue 11
Keywords 损失函数
SimAM
aluminum profiles
loss function
distillation
量化
蒸馏
defects detection
铝型材
缺陷检测
quantization
Language Chinese
LinkModel OpenURL
MergedId FETCHMERGED-LOGICAL-s1025-229ae602bd4a0e99c87082444b0191f15e69525250c35f659cf2bce72d4e8c4b3
PageCount 8
ParticipantIDs wanfang_journals_dbdxxb202411005
PublicationCentury 2000
PublicationDate 2024-11-15
PublicationDateYYYYMMDD 2024-11-15
PublicationDate_xml – month: 11
  year: 2024
  text: 2024-11-15
  day: 15
PublicationDecade 2020
PublicationTitle 东北大学学报（自然科学版）
PublicationTitle_FL Journal of Northeastern University(Natural Science)
PublicationYear 2024
Publisher 东北大学秦皇岛分校 控制工程学院,河北 秦皇岛 066004
Publisher_xml – name: 东北大学秦皇岛分校 控制工程学院,河北 秦皇岛 066004
SSID ssib000947529
ssib051368049
ssib023167010
ssj0040330
ssib002039846
ssib004675270
ssib006703041
ssib002263414
ssib008679651
ssib001128993
Score 2.4100776
Snippet TP391.4; 铝型材在生产加工过程中会产生特征不明显和尺度大小不一等多类型的表面缺陷,针对现有人工抽检方法准确率低、实时性差、主观性强等问题,提出一种基于PP-PicoDet-XS的改进铝型材表面缺陷检测算法.改进的算法在主干网络中嵌入无参注意力SimAM,增强对深层有效特征的提取能力;使用SIoU(Scylla...
SourceID wanfang
SourceType Aggregation Database
StartPage 1557
Title 基于PP-PicoDet-XS的改进铝型材表面缺陷检测算法
URI https://d.wanfangdata.com.cn/periodical/dbdxxb202411005
Volume 45
hasFullText 1
inHoldings 1
isFullTextHit
isPrint
link http://utb.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwnV3Na9RAFA_9ANGD-Inf9OA7lazJJDOZOXiY2U0oglKwhd7KJpvV0wp2C6UnD94KIqiIShX14smLiFgE_xm7bf8HD743ye6OUmoVljDMvHnvzXubeb9JZl487yrjJS9FO_dLXFD7cRlIvy3y0BdJqVhH5nGVYuPmLTG3GN9Y4ksTkz-dXUur_bxRrO97ruR_vIp16Fc6JfsPnh0xxQoso3_xih7G66F8DCkHlYHRkMZ0len8vD-Pvm2VfX_pNqQJKKyNIRWgkEJBKsFkoAykClQEqmVZpCCNpWmBCohGh6ClpWmBZsTHNK0YrFFgEiLWEe2SwILhtnsCOgVlmwxy5i7stfpJUE0SJ5uWjIOOQVcFlCKcggCpQXNIM5IrJakkkb8mKTgcI6w4HF3o9MImZYmrXqMnHKS1xl_TqjbsJFtorjGJHbxMLV9TmwMV1dmYxEqs1DMhyaJCACp0uRhGFiJ21sAoGq0uM_fhCovplGF1vNTeDoezzkh85VTbRNKM7YUGsprpAJ03a7VP6_7YZqx1cexk1IR8K43D2noV_cya9RDQgUNVZg-QPItIMqi-61wHNso4GwVVdoJh5KsSeQ7v8NCJY4gyEwcTUUahfeMtC4S0AZeENEZCGmTLBmXHDfgYaIy2f3byztpaTjSUsJBPetMsSWiTxbQ2LZM58D9OuAPfcW0glQtPg0hJNz0eEwjK3HPX2N2JR4LCmwP_bbLJ8VtxRlkhgjEc5mEkpH0LXSG7OIiiKrtJPcwjHgyNcO0gE9gDgb1uu3fHwa4LJ7zj9aJzRlczyElvYv3uKe-Yk4r0tHd9-83Wj61Hv80fOy8fDp5-3f3-au_J5vbrjcHm4923H_Y23-1829p78WXw_sHg88bOx-eDT8_OeItZutCc8-tPq_grIX3AmjHVLkXA8k7cDkqlCgzbEpF-nOOSL-yGOH8rzmjLQxHxruCq6LK8KBPWiUtZxHl01pvq3euV57yZJIzaAUJcRYfacf2Zs5IVLFKhRGYqCc97M_Xgl-upc2X5jz_Ahb-TXPSOju_PS95U__5qeRmXA_38Sv2v-QXGZrtO
linkProvider EBSCOhost
openUrl ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Ajournal&rft.genre=article&rft.atitle=%E5%9F%BA%E4%BA%8EPP-PicoDet-XS%E7%9A%84%E6%94%B9%E8%BF%9B%E9%93%9D%E5%9E%8B%E6%9D%90%E8%A1%A8%E9%9D%A2%E7%BC%BA%E9%99%B7%E6%A3%80%E6%B5%8B%E7%AE%97%E6%B3%95&rft.jtitle=%E4%B8%9C%E5%8C%97%E5%A4%A7%E5%AD%A6%E5%AD%A6%E6%8A%A5%EF%BC%88%E8%87%AA%E7%84%B6%E7%A7%91%E5%AD%A6%E7%89%88%EF%BC%89&rft.au=%E9%A9%AC%E6%B7%91%E5%8D%8E&rft.au=%E6%9D%8E%E7%AB%8B%E6%8C%AF&rft.au=%E7%A7%A6%E6%B1%89%E6%B0%91&rft.au=%E6%B2%99%E6%99%93%E9%B9%8F&rft.date=2024-11-15&rft.pub=%E4%B8%9C%E5%8C%97%E5%A4%A7%E5%AD%A6%E7%A7%A6%E7%9A%87%E5%B2%9B%E5%88%86%E6%A0%A1+%E6%8E%A7%E5%88%B6%E5%B7%A5%E7%A8%8B%E5%AD%A6%E9%99%A2%2C%E6%B2%B3%E5%8C%97+%E7%A7%A6%E7%9A%87%E5%B2%9B+066004&rft.issn=1005-3026&rft.volume=45&rft.issue=11&rft.spage=1557&rft.epage=1564&rft_id=info:doi/10.12068%2Fj.issn.1005-3026.2024.11.005&rft.externalDocID=dbdxxb202411005
thumbnail_s http://utb.summon.serialssolutions.com/2.0.0/image/custom?url=http%3A%2F%2Fwww.wanfangdata.com.cn%2Fimages%2FPeriodicalImages%2Fdbdxxb%2Fdbdxxb.jpg