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

相关实验视频

Updated: May 9, 2025

Deep Neural Networks for Image-Based Dietary Assessment
13:19

Deep Neural Networks for Image-Based Dietary Assessment

Published on: March 13, 2021

8.9K

通过噪声注入来学习最佳图像表示,用于细粒度搜索.

Vidit Kumar1, Vikas Tripathi1, Bhaskar Pant1

  • 1Department of CSE, Graphic Era Deemed to be University, Dehradun, India.

Scientific reports
|May 3, 2025
PubMed
概括

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

Correction: Multimodality medical image fusion using directional total variation based linear spectral clustering in NSCT domain.

Scientific reports·2026
Same author

A switching-based deep learning framework for personalized and adaptive E-commerce recommendations.

Scientific reports·2026
Same author

Multimodality medical image fusion using directional total variation based linear spectral clustering in NSCT domain.

Scientific reports·2026
Same author

A hybrid model for improved nail disease classification using vision transformers and stable diffusion.

Scientific reports·2026
Same author

The RNA-binding protein Quaking is essential for cardiac homeostasis and function by regulating Morf4l2 splicing.

Journal of molecular and cellular cardiology·2025
Same author

Intelligent traffic light management using predictive and dynamic traffic flow analysis.

Scientific reports·2025

这项研究引入了一种新的噪音注入方法,以改善细粒度图像搜索. 通过向图像和特征添加噪音,该方法增强了表示学习,并实现了卓越的检索结果.

科学领域:

  • 计算机视觉 计算机视觉
  • 机器学习 机器学习

背景情况:

  • 细粒度图像搜索是一个日益兴趣的领域.
  • 当前的方法通常使用嵌入式的深度功能学习,但面临诸如昂贵的采样 (三倍损失) 或早期和 (软max损失) 等挑战.

研究的目的:

  • 为了增强用于图像搜索的细粒度表示学习.
  • 为了解决现有的损失函数的局限性,如三倍损失和软max损失.

主要方法:

  • 一种新的方法,将噪音注入到输入图像和深度特征中.
  • 缩小L2在嵌入空间中的原始图像和噪音图像的规范特征之间的距离.
  • 使用特征噪声作为规范化,以防止过拟合和促进通用特征.

主要成果:

  • 与现有方法相比,在牛津花-17,Cub-200-2011和Cars-196数据集上取得了优异的检索结果.
  • 在汽车-196和Cub-200-2011的零射击设置中表现出良好的性能.

结论:

  • 拟议的噪音注入技术有效地增强了细粒度表示学习.
  • 这种方法为改善图像搜索准确性和概括性提供了一个有希望的替代方案.
关键词:
功能学习的特点是:精细粒度图像检索 精细粒度图像检索图像表示图像表示.噪音注入的噪音注入零射击学习的学习.

更多相关视频

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

428
Integration of Animal Behavioral Assessment and Convolutional Neural Network to Study Wasabi-Alcohol Taste-Smell Interaction
06:19

Integration of Animal Behavioral Assessment and Convolutional Neural Network to Study Wasabi-Alcohol Taste-Smell Interaction

Published on: August 16, 2024

346

相关实验视频

Last Updated: May 9, 2025

Deep Neural Networks for Image-Based Dietary Assessment
13:19

Deep Neural Networks for Image-Based Dietary Assessment

Published on: March 13, 2021

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

428
Integration of Animal Behavioral Assessment and Convolutional Neural Network to Study Wasabi-Alcohol Taste-Smell Interaction
06:19

Integration of Animal Behavioral Assessment and Convolutional Neural Network to Study Wasabi-Alcohol Taste-Smell Interaction

Published on: August 16, 2024

346