Jove
Visualize
联系我们
JoVE
x logofacebook logolinkedin logoyoutube logo
关于 JoVE
概览领导团队博客JoVE 帮助中心
作者
出版流程编辑委员会范围与政策同行评审常见问题投稿
图书馆员
用户评价订阅访问资源图书馆顾问委员会常见问题
研究
JoVE JournalMethods CollectionsJoVE Encyclopedia of Experiments存档
教育
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab Manual教师资源中心教师网站
使用条款与条件
隐私政策
政策

相关概念视频

Detection of Black Holes01:10

Detection of Black Holes

Although black holes were theoretically postulated in the 1920s, they remained outside the domain of observational astronomy until the 1970s.
Their closest cousins are neutron stars, which are composed almost entirely of neutrons packed against each other, making them extremely dense. A neutron star has the same mass as the Sun but its diameter is only a few kilometers. Therefore, the escape velocity from their surface is close to the speed of light.
Not until the 1960s, when the first neutron...

您也可能阅读

相关文章

通过共同作者、期刊和引用图与本文相关的文章。

排序
Same author

Iridocorneal endothelial syndrome manifesting 4.5 years after toric implantable collamer lens implantation: a case report.

BMC ophthalmology·2026
Same author

Bioinspired Carbon Radical Catalysis.

Journal of the American Chemical Society·2026
Same author

DHCR7 promotes granulosa cells ferroptosis and represents a potential therapeutic target in premature ovarian insufficiency.

Biochimica et biophysica acta. Molecular cell research·2026
Same author

Proteomic and transcriptomic analyses identify the role of RBM25 in the malignant progression of glioma.

The International journal of neuroscience·2026
Same author

Association of gut microbiota and inflammatory markers with enteral nutrition intolerance in patients with early-stage moderate-to-severe intracerebral hemorrhage.

Microbiology spectrum·2026
Same author

Multi-Stage State Assessment of Breakers Based on TCWGAN-GP and XGBoost Under Insufficient Samples.

Sensors (Basel, Switzerland)·2026

相关实验视频

Updated: Jul 10, 2026

Subsurface Defect Localization by Structured Heating Using Laser Projected Photothermal Thermography
11:34

Subsurface Defect Localization by Structured Heating Using Laser Projected Photothermal Thermography

Published on: May 15, 2017

11.1K

基于YOLO接的激光接点检测.

Jianxin Feng1,2, Jiahao Wang3,4, Xinyu Zhao3,4

  • 1Communication and Network Laboratory, Dalian University, Dalian, 116622, China. fengjianxin863@163.com.

Scientific reports
|November 26, 2024
PubMed
概括

本研究介绍了YOLO-Weld,这是激光接点检测的高效模型. 它提高了准确性并减少了参数,解决了在工业制造中数据有限和不规则形状等挑战.

关键词:
边界框回归的边界框回归方法激光接点检测 激光接点检测多个尺度的特征多个尺度的特征.长尾效应是一种长尾效应.通过YOLO接来进行接.

更多相关视频

Quantitative Analysis of Vacuum Induction Melting by Laser-induced Breakdown Spectroscopy
03:49

Quantitative Analysis of Vacuum Induction Melting by Laser-induced Breakdown Spectroscopy

Published on: June 10, 2019

7.2K
Generating Lap Joints Via Friction Stir Spot Welding on DP780 Steel
07:18

Generating Lap Joints Via Friction Stir Spot Welding on DP780 Steel

Published on: August 13, 2019

6.9K

相关实验视频

Last Updated: Jul 10, 2026

Subsurface Defect Localization by Structured Heating Using Laser Projected Photothermal Thermography
11:34

Subsurface Defect Localization by Structured Heating Using Laser Projected Photothermal Thermography

Published on: May 15, 2017

11.1K
Quantitative Analysis of Vacuum Induction Melting by Laser-induced Breakdown Spectroscopy
03:49

Quantitative Analysis of Vacuum Induction Melting by Laser-induced Breakdown Spectroscopy

Published on: June 10, 2019

7.2K
Generating Lap Joints Via Friction Stir Spot Welding on DP780 Steel
07:18

Generating Lap Joints Via Friction Stir Spot Welding on DP780 Steel

Published on: August 13, 2019

6.9K

科学领域:

  • 工业制造业 工业制造业 工业制造业
  • 计算机视觉 计算机视觉
  • 机器学习 机器学习

背景情况:

  • 激光接点检测至关重要,但受到有限,分布不均,形状不规则的样本的挑战.
  • 现有的方法在数据稀缺和对象外观多样性方面扎.

研究的目的:

  • 开发一种创新的轻型模型 (YOLO-Weld),以提高激光接点检测的准确性和效率.
  • 在具有挑战性的工业场景中解决数据不平衡并改善边界框回归.

主要方法:

  • 雇佣针对少数群体的数据增强,并引入多元类正常化损失 (DCNLoss) 以优先考虑尾部数据.
  • 在联盟损失 (AHIoU损失) 上开发了自适应层次交叉,以专注于中度IOU样本,加速回归.
  • 提出了一个轻量级的多尺度特征处理模块MSBCSPELAN,以优化特征处理和减少模型大小.

主要成果:

  • YOLO-Weld显著提高了检测准确度,mAP@50增加了15.6%,mAP@50:95增加了15.8%.
  • 该模型的精度提高了4.3%,回忆率提高了22.2%,F1分数增加了15.1%.
  • 参数数量减少了0.4M,GFLOPS减少了1.1,这表明模型更有效.

结论:

  • YOLO-Weld为激光接点检测提供了卓越的解决方案,克服了数据限制并提高了性能.
  • 拟议的模型证明了工业应用的提高准确性,效率和稳定性.