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

Artificial Intelligence for Evidence Synthesis of Emerging Biologics to Improve Skeletal Health in Osteogenesis Imperfecta: Systematic Review and Meta-Analysis.

Journal of medical Internet research·2026
Same author

Extracting patient reported cannabis use and reasons for use from electronic health records: a benchmarking study of large language models.

medRxiv : the preprint server for health sciences·2026
Same author

Incidence of cutaneous lupus erythematosus and systemic progression: A Nationwide study in Korea.

Journal of the European Academy of Dermatology and Venereology : JEADV·2026
Same author

Incidence and risk factors of first-ever diabetic foot ulcers in patients with newly diagnosed type 2 diabetes mellitus.

Diabetes research and clinical practice·2026
Same author

Machine learning and deep learning-based drug-drug interactions prediction: a systematic review focused on anticancer drugs.

NPJ precision oncology·2026
Same author

Associations between remnant cholesterol and cardiovascular events in patients undergoing dialysis: a nationwide cohort study.

Kidney research and clinical practice·2026

相关实验视频

Updated: May 24, 2025

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

451

通过知识蒸重新审视域适应性语义细分.

Seongwon Jeong, Jiyeong Kim, Sungheui Kim

    IEEE transactions on image processing : a publication of the IEEE Signal Processing Society
    |March 3, 2025
    PubMed
    概括

    本研究引入了一种新的方法,通过使用两个教师模型在语义细分中进行无监督域调整. 这种方法增强了伪地面真相生成和知识传输,提高了细分的准确性.

    科学领域:

    • 计算机视觉 计算机视觉
    • 机器学习 机器学习
    • 深度学习 (Deep Learning) 是一种深度学习.

    背景情况:

    • 无监督域调整 (UDA) 对于语义细分至关重要,自监督的方法显示出有前途.
    • 现有的自我训练UDA方法有风险传播从伪基础真理 (PGT) 到教师模型的不准确性.
    • 教师模型中的指数移动平均值 (EMA) 更新仍然可能导致错误传播.

    研究的目的:

    • 提出一种新的UDA方法用于语义细分,使用两个不同的教师模型.
    • 为解决不准确的PGT和教师模型更新在自培训UDA中的问题.
    • 为了利用知识蒸 (KD) 原则来改善语义细分中的UDA.

    主要方法:

    • 一个新的UDA方法,采用两种教师模型:一个是EMA更新的PGT生成,另一个是结的,预先训练的教师,用于特征空间知识传递.
    • 利用一个结的教师模型,具有潜在的更大的代表权力,不受架构的约束.
    • 从KD的角度重新审视自我训练的UDA.

    主要成果:

    • 拟议的方法增强了目标领域的语义细分性能,跨越各种骨干和场景.
    • 在不同的实验设置中证明了可扩展性 (GTA5 → Cityscapes,SYNTHIA → Cityscapes).

    更多相关视频

    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

    2.6K
    From Voxels to Knowledge: A Practical Guide to the Segmentation of Complex Electron Microscopy 3D-Data
    12:08

    From Voxels to Knowledge: A Practical Guide to the Segmentation of Complex Electron Microscopy 3D-Data

    Published on: August 13, 2014

    24.5K

    相关实验视频

    Last Updated: May 24, 2025

    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

    451
    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

    2.6K
    From Voxels to Knowledge: A Practical Guide to the Segmentation of Complex Electron Microscopy 3D-Data
    12:08

    From Voxels to Knowledge: A Practical Guide to the Segmentation of Complex Electron Microscopy 3D-Data

    Published on: August 13, 2014

    24.5K
  • 实现与最先进的方法相比或优于其性能,即使使用轻量级的骨架.

    • 结论:

    • 双教师模型方法有效地减轻了PGT中的不准确性,并改善了教师模型的更新.
    • 这种方法为语义细分中的无监督域调整提供了一个可扩展和强大的解决方案.
    • 拟议的技术展示了KD启发的战略在推进UDA研究方面的潜力.