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

Thymosin Alpha-1 Restores Chemotherapy-Induced Antitumor Immunity by Chaperoning a MicroRNA Ligand of TLR7 in Dendritic Cells.

Cancer research·2026
Same author

PMF-YOLO: a parameter-efficient multi-scale detector for underwater optical imagery under scattering and nonuniform illumination.

Applied optics·2026
Same author

Dual-branch underwater image enhancement network based on fusion of polarization and color information.

Applied optics·2026
Same author

Mechanistic insights into NaOH-driven crystal destruction and selective lithium extraction from spent LiFePO4 via mechanochemistry.

Waste management (New York, N.Y.)·2026
Same author

On-site monitoring of juice spoilage: a POCT immunosensor based on sponge-like sensing interface for alicyclobacillus acidoterrestris detection.

Food research international (Ottawa, Ont.)·2026
Same author

Metal-Phenolic Networks Delay the Oxidation of Alkaline High-Protein Gel Foods: Improving the Quality of Coated Tofu.

Gels (Basel, Switzerland)·2026

相关实验视频

Updated: Jul 23, 2025

Estimation of Contact Regions Between Hands and Objects During Human Multi-Digit Grasping
09:41

Estimation of Contact Regions Between Hands and Objects During Human Multi-Digit Grasping

Published on: April 21, 2023

1.6K

基于转移学习的小型样本多目标掌握技术的研究.

Bin Zhao1,2,3, Chengdong Wu3, Fengshan Zou2

  • 1College of Information Science and Engineering, Northeastern University, Shenyang 110819, China.

Sensors (Basel, Switzerland)
|July 14, 2023
PubMed
概括

这项研究引入了一种新的CBAM-ASPP-SqueezeNet模型,用于机器人多目标抓取检测. 该模型在物理实验中取得了93%的成功率,增强了机器人操纵能力.

关键词:
这就是SqueezeNet.注意力机制注意力机制深度学习是一种深度学习.抓住 检测 抓住 检测多对象检测多对象检测

更多相关视频

Design and Use of an Apparatus for Presenting Graspable Objects in 3D Workspace
09:11

Design and Use of an Apparatus for Presenting Graspable Objects in 3D Workspace

Published on: August 8, 2019

5.8K
Author Spotlight: Enhancing Grasping Abilities for Hemiplegic Patients with Flexible Robotic Limbs
03:55

Author Spotlight: Enhancing Grasping Abilities for Hemiplegic Patients with Flexible Robotic Limbs

Published on: October 27, 2023

2.2K

相关实验视频

Last Updated: Jul 23, 2025

Estimation of Contact Regions Between Hands and Objects During Human Multi-Digit Grasping
09:41

Estimation of Contact Regions Between Hands and Objects During Human Multi-Digit Grasping

Published on: April 21, 2023

1.6K
Design and Use of an Apparatus for Presenting Graspable Objects in 3D Workspace
09:11

Design and Use of an Apparatus for Presenting Graspable Objects in 3D Workspace

Published on: August 8, 2019

5.8K
Author Spotlight: Enhancing Grasping Abilities for Hemiplegic Patients with Flexible Robotic Limbs
03:55

Author Spotlight: Enhancing Grasping Abilities for Hemiplegic Patients with Flexible Robotic Limbs

Published on: October 27, 2023

2.2K

科学领域:

  • 机器人技术 机器人技术 机器人技术
  • 计算机视觉 计算机视觉
  • 机器学习 机器学习

背景情况:

  • 机器人多目标抓取检测对于先进的自动化至关重要.
  • 现有的方法在准确识别和定位多个物体以抓取时面临挑战.
  • 注意力机制和空间金字塔聚合的整合为改善特征提取提供了潜力.

研究的目的:

  • 提出和验证一个新的模型,CBAM-ASPP-SqueezeNet,用于增强机器人多目标抓取检测.
  • 为了利用转移学习,在定制的多目标掌握数据集上进行高效的模型培训.
  • 改进SqueezeNet的架构,使用道和空间注意力与的空间金字塔聚合.

主要方法:

  • 开发和扩展一个多目标抓取数据集.
  • 转移学习的应用在单一目标数据集上进行预培训.
  • 优化SqueezeNet模型与卷积块注意模块 (CBAM) 和形空间金字塔聚合 (ASPP).
  • 在频道和空间维度中实现特征图权重的注意力机制.
  • 利用不同速率的状卷曲来扩大受体场并捕捉多尺度的特征.

主要成果:

  • 通过CBAM-ASPP-SqueezeNet模型,经过20个培训时代,与转移学习进行了融合.
  • 在使用Kinova和SIASUN机器人手臂的物理实验中,获得了93%的显著抓取成功率.
  • 提出的注意力和ASPP模块有效地改善了掌握任务的特征表示.

结论:

  • 该CBAM-ASPP-SqueezeNet模型是有效的机器人多目标抓住检测.
  • 转移学习加速了融合,并改善了定制数据集的性能.
  • 注意力机制和ASPP的整合提高了机器人系统在复杂的操纵任务中的能力.