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

Aluminum-Triggered Dealloying for Hierarchical Porous High-Entropy Alloy Electrodes Enabling Industrially Stable Oxygen Evolution.

ACS nano·2026
Same author

Hierarchical Porous Collectors with Lithiophilicity and Conductivity Gradients for Regulated Lithium Deposition and Stable Lithium Metal Anodes.

Small (Weinheim an der Bergstrasse, Germany)·2026
Same author

Constructing Reaction-Limited and Robust Interface in 2D MAB-Phase MoAlB Nanosheet-Reinforced Copper Matrix Composites.

ACS applied materials & interfaces·2026
Same author

Soluble-Salt-Template Synthesis of Nanosheets With Tunable Two-/Three-Dimensional Architectures for Electrochemical Energy Storage and Conversion.

Angewandte Chemie (International ed. in English)·2026
Same author

Addressing Oxygen Embrittlement in Additively Manufactured Titanium via Cu-Mediated Interstitial Site Engineering.

Advanced science (Weinheim, Baden-Wurttemberg, Germany)·2026
Same author

Aggregation kinetics and stability of Ce(IV)-borate colloid: Effect of pH, electrolytes, gibbsite colloid, and humic acid colloid.

Journal of contaminant hydrology·2025
Same journal

HardFlow: Hard-Constrained Sampling for Flow-Matching Models Via Trajectory Optimization.

IEEE transactions on pattern analysis and machine intelligence·2026
Same journal

Industrial Brain: Self-Evolving Neuro-Symbolic Autonomy with Causal Resilience for Cyber-Physical Systems.

IEEE transactions on pattern analysis and machine intelligence·2026
Same journal

Adaptive Hardness-Driven Dictionary Distillation for Incomplete Streaming View Clustering.

IEEE transactions on pattern analysis and machine intelligence·2026
Same journal

Mixture of Global and Local Experts with Diffusion Transformer for Controllable Face Generation.

IEEE transactions on pattern analysis and machine intelligence·2026
Same journal

Task-KV: Task-aware KV Cache Optimization via Semantic Differentiation of Attention Heads.

IEEE transactions on pattern analysis and machine intelligence·2026
Same journal

Achieving Text-based Person Retrieval with Any Granularity.

IEEE transactions on pattern analysis and machine intelligence·2026
查看所有相关文章

相关实验视频

Updated: Jan 17, 2026

Preparation and Applications of Organotypic Thymic Slice Cultures
10:10

Preparation and Applications of Organotypic Thymic Slice Cultures

Published on: August 6, 2016

8.7K

SOOD++:利用未标记的数据来增强面向对象检测

Dingkang Liang, Wei Hua, Chunsheng Shi

    IEEE transactions on pattern analysis and machine intelligence
    |September 18, 2025
    PubMed
    概括
    此摘要是机器生成的。

    半监控定向物体检测 (SSOD) 通过SOOD++进行了改进,它有效地利用未标记的空中图像. 这种方法产生了密集的伪标签,并使用几何感知损失来改进对定向对象的检测.

    更多相关视频

    Differentiation of Functional Osteoclasts from Human Peripheral Blood CD14+ Monocytes
    11:52

    Differentiation of Functional Osteoclasts from Human Peripheral Blood CD14+ Monocytes

    Published on: January 27, 2023

    4.3K
    Engineered 3D Silk-collagen-based Model of Polarized Neural Tissue
    06:17

    Engineered 3D Silk-collagen-based Model of Polarized Neural Tissue

    Published on: October 23, 2015

    12.9K

    相关实验视频

    Last Updated: Jan 17, 2026

    Preparation and Applications of Organotypic Thymic Slice Cultures
    10:10

    Preparation and Applications of Organotypic Thymic Slice Cultures

    Published on: August 6, 2016

    8.7K
    Differentiation of Functional Osteoclasts from Human Peripheral Blood CD14+ Monocytes
    11:52

    Differentiation of Functional Osteoclasts from Human Peripheral Blood CD14+ Monocytes

    Published on: January 27, 2023

    4.3K
    Engineered 3D Silk-collagen-based Model of Polarized Neural Tissue
    06:17

    Engineered 3D Silk-collagen-based Model of Polarized Neural Tissue

    Published on: October 23, 2015

    12.9K

    科学领域:

    • 计算机视觉 计算机视觉
    • 机器学习 机器学习
    • 遥感 遥感 遥感 遥感

    背景情况:

    • 半监控物体检测 (SSOD) 通常侧重于水平物体,忽视空中图像中普遍存在的定向物体.
    • 面向对象的高注释成本在开发有效的检测模型方面构成了重大挑战.
    • 现有的SSOD方法不能充分解决空中物体的独特特征,例如任意的方向,小尺度和密集的分布.

    研究的目的:

    • 提出一种新的半监督定向物体检测 (SOOD++) 方法,以解决空中成像当前SSOD方法的局限性.
    • 开发一种技术,有效地利用未标记的数据来更好地检测定向对象,减少对昂贵注释的依赖.
    • 通过半监督学习,在面向对象检测方面建立一个新的最先进的技术.

    主要方法:

    • 一个简单的实例意识密集采样 (SIDS) 策略被用来生成面向对象的全面密集伪标签.
    • 具有几何意识的自适应权衡 (GAW) 损失动态调整基于空中物体几何信息的预测的重要性.
    • 噪音驱动的全球一致性 (NGC) 明确模拟伪标签和预测之间的多对多关系,将空中图像视为全球布局.

    主要成果:

    • 在各种标记数据设置 (10%,20%,30%) 中,SOOD++在DOTA-V2.0/DOTA-V1.5基准测试上显著优于之前的先进方法.
    • 该方法实现了显著的mAP改进,例如,在DOTA-V2.0上+2.90 mAP,标记数据为10%.
    • SOOD++超过了强大的监督基线+1.82 mAP,在DOTA-V1.5.5.上达到72.48 mAP的最新状态.
    • 在不同的定向探测器中表现出稳定的泛化,包括多视图定向的3D对象探测器.

    结论:

    • 提出的SOOD++方法是有效和简单的,用于在空中图像中进行半监督的定向物体检测.
    • SOOD++成功地解决了空中物体检测中任意方向,小尺度和密集分布的挑战.
    • 该方法提供了显著的性能提升,并建立了一个新的最先进的状态,突出了半监督学习在这个领域的潜力.