Search research articles

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 Journal Methods Collections JoVE Encyclopedia of Experiments 存档

教育

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual 教师资源中心教师网站

使用条款与条件

相关概念视频

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

RICNET: Retinex-Inspired Illumination Curve Estimation for Low-Light Enhancement in Industrial Welding Scenes.

Sensors (Basel, Switzerland)·2025

Same author

A Welding Defect Detection Model Based on Hybrid-Enhanced Multi-Granularity Spatiotemporal Representation Learning.

Sensors (Basel, Switzerland)·2025

Same author

Transcriptome and proteome of ovarian tissues revealed the differences in the ovaries of dairy goats during the breeding and non-breeding seasons.

Frontiers in veterinary science·2025

Same author

Predictors of sudden cardiac death in postmyectomy hypertrophic cardiomyopathy with obstruction.

Heart rhythm·2025

Same author

CD36 molecule and AMP-activated protein kinase signaling drive docosahexaenoic acid-induced lipid remodeling in goat mammary epithelial cells.

International journal of biological macromolecules·2025

Same author

Predictors of N-Terminal Pro-Brain Natriuretic Peptide Nonresponse and Its Impact on Outcomes in Patients After Septal Myectomy.

The Annals of thoracic surgery·2025

Search research articles

相关实验视频

Updated: Jun 12, 2025

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Published on: November 30, 2022

对工业表面缺陷进行高效准确的半监督语义细分.

Chenbo Shi^1,2, Kang Wang¹, Guodong Zhang²

¹College of Intelligent Equipment, Shandong University of Science and Technology, Taian, 271019, China.

Scientific reports

|September 19, 2024

概括

此摘要是机器生成的。

本研究引入了一种新的半监督语义细分框架,用于工业缺陷检测. 它通过使用扰动不变性和多种交叉伪监督来提高准确性和速度,克服了当前深度学习方法的局限性.

关键词:

深度神经网络是一个神经网络.缺陷检测检测缺陷检测检测缺陷检测工业化工业化工业化工业化工业化分段化分段化分段化分段化半监督学习半监督学习

更多相关视频

Author Spotlight: Efficient Image Recognition Using Directional Gradient Histogram Technique and Support Vector Machines

Author Spotlight: Efficient Image Recognition Using Directional Gradient Histogram Technique and Support Vector Machines

Published on: January 5, 2024

Combining Eye-tracking Data with an Analysis of Video Content from Free-viewing a Video of a Walk in an Urban Park Environment

Combining Eye-tracking Data with an Analysis of Video Content from Free-viewing a Video of a Walk in an Urban Park Environment

Published on: May 7, 2019

相关实验视频

Last Updated: Jun 12, 2025

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Published on: November 30, 2022

Author Spotlight: Efficient Image Recognition Using Directional Gradient Histogram Technique and Support Vector Machines

Author Spotlight: Efficient Image Recognition Using Directional Gradient Histogram Technique and Support Vector Machines

Published on: January 5, 2024

Combining Eye-tracking Data with an Analysis of Video Content from Free-viewing a Video of a Walk in an Urban Park Environment

Combining Eye-tracking Data with an Analysis of Video Content from Free-viewing a Video of a Walk in an Urban Park Environment

Published on: May 7, 2019

科学领域:

计算机视觉计算机视觉
机器学习机器学习
工业自动化工业自动化

背景情况:

深度学习方法对于工业质量检查至关重要,但面临着有限数据,低利用率和性能瓶的挑战.
现有的半监督方法在复杂的工业缺陷检测场景中难以准确和快速.

研究的目的:

提出一个新的半监督语义细分框架,以解决工业缺陷检测的局限性.
在自动化质量检查中提高准确性,数据利用率和实时性能.

主要方法:

开发了一个半监督的语义细分框架,利用图像和特征空间中的扰动不变性.
采用多种扰动交叉伪监控来最大限度地减少对大型标记数据集的依赖.
集成边缘像素级语义信息和浅功能融合,以提高效率和准确性.

主要成果:

拟议的方法在工业环境中明显优于最先进的 (SOTA) 半监督语义细分技术.
在定制数据集上实现了比SOTA高3.11%的平均交叉点在整个欧盟 (mIoU),在KolektorSDD数据集上高4.39%.
该网络与U-net的速度相匹配,同时在mIoU中超过DeepLabv3Plus的2.99%.

结论:

开发的框架提供了一个强大的和有效的解决方案,用于工业缺陷检测使用半监督学习.
该方法有效地处理了样本大小不足,并改善了小目标和缺陷边缘的检测.
这种方法通过提高准确性和实时处理能力来推进自动化质量检查.