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

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

Comprehensive analysis of the multi-rings mitochondrial genome of Populus tomentosa.

BMC genomics·2025
Same author

Evaluation of Cold Resistance in Alfalfa Varieties Based on Root Traits and Winter Survival in Horqin Sandy Land.

Biology·2025
Same author

Tryptophan metabolism-related gene CYP1B1 serves as a shared biomarker for both Parkinson's disease and insomnia.

Scientific reports·2025
Same author

A study of specific immunoglobulin G4 expression in allergic rhinitis and its value in assessing efficacy and in predicting prognosis of sublingual immunotherapy.

The Kaohsiung journal of medical sciences·2024
Same author

Non-affinity platform for processing knob-into-hole bispecific antibody.

Bioresources and bioprocessing·2024
Same author

High Relative Humidity-Induced Growth of Perovskite Nanowires from Glass toward Single-Mode Photonic Nanolasers at Sub-100-nm Scale.

Advanced science (Weinheim, Baden-Wurttemberg, Germany)·2024

相关实验视频

Updated: Jun 9, 2025

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
04:48

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

Published on: November 30, 2022

2.7K

使用图像级标签的一步三倍增强弱监督的语义细分.

Longjie Quan1, Dandan Huang1, Zhi Liu1,2

  • 1School of Electronics and Information Engineering, Changchun University of Science and Technology, Changchun, People's Republic of China.

PloS one
|October 21, 2024
PubMed
概括

本研究引入了一步三倍增强 (OSTE) 网络,用于弱监督的语义细分,简化模型并使用图像级标签提高准确性. 通过增强本地化和完善细分边界,OSTE比现有方法取得更好的结果.

更多相关视频

Combining Eye-tracking Data with an Analysis of Video Content from Free-viewing a Video of a Walk in an Urban Park Environment
08:25

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

8.9K
Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications
03:31

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications

Published on: December 15, 2023

475

相关实验视频

Last Updated: Jun 9, 2025

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
04:48

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

Published on: November 30, 2022

2.7K
Combining Eye-tracking Data with an Analysis of Video Content from Free-viewing a Video of a Walk in an Urban Park Environment
08:25

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

8.9K
Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications
03:31

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications

Published on: December 15, 2023

475

科学领域:

  • 计算机科学 计算机科学
  • 人工智能的人工智能
  • 机器学习 机器学习

背景情况:

  • 传统的语义细分需要像素级标签,这是耗时和资源密集的.
  • 弱监督的语义细分使用图像级标签,降低注释成本,但往往缺乏精确的本地化信息.
  • 现有的监管较弱的方法通常采用复杂的两步方法,增加模型参数和结构复杂性.

研究的目的:

  • 提出创新的一步三倍增强 (OSTE) 弱监督的语义细分网络.
  • 通过将伪标签生成和细分集成到单个步骤中来简化模型结构.
  • 通过使用图像级标签来提高本地化和边界精细化来提高细分精度.

主要方法:

  • 开发了OSTE网络,用于伪标签生成和语义细分的一步方法.
  • 集成的本地激活地图信息与图像,以改善本地化和扩展功能.
  • 通过利用多层次的特征相关性,精细的类激活地图种子区域.
  • 包含条件随机场理论,用于生成具有丰富边界细节的高可靠性伪标签.

主要成果:

  • 在帕斯卡的VOC数据集中,OSTE网络在欧盟 (mIoU) 上获得了58.47%的竞争平均交叉点.
  • 与当前的两步弱监督的语义细分方案相比,显著改进.
  • 在mIoU得分中,超过现有的端到端计划的表现至少为5.03%.

结论:

  • 拟议的OSTE网络提供了一种更简单,更有效的方法来对弱监督的语义细分.
  • 三重增强策略显著提高了细分精度,特别是在本地化和边界定义方面.
  • OSTE为现有方法提供了有竞争力的替代方案,减少了复杂性,同时实现了卓越的性能.