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

A Tetrameric Biotin-Avidin Scaffold Using Anti-CD19 (FMC63) Traptavidin Fusion Enables High-Avidity CAR-T Cells for Leukemia Therapy.

Bioconjugate chemistry·2025
Same author

Development of antibodies against severe fever with thrombocytopenia syndrome virus nucleoprotein for diagnosis.

Applied microbiology and biotechnology·2025
Same author

Unveiling CNS cell morphology with deep learning: A gateway to anti-inflammatory compound screening.

PloS one·2025
Same author

Mechanistic Fatigue Performance Evaluation of Stone Mastic Asphalt Mixtures: Effect of Asphalt Performance Grade and Elastic Recovery.

Polymers·2024
Same author

Synthetic Data Enhancement and Network Compression Technology of Monocular Depth Estimation for Real-Time Autonomous Driving System.

Sensors (Basel, Switzerland)·2024
Same author

Photoprotective effects of Lithospermum erythrorhizon and Pueraria lobata extracts on UVB-induced photoaging: A study on skin barrier protection.

Photodermatology, photoimmunology & photomedicine·2024

相关实验视频

Updated: May 14, 2025

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

基于摄像头的BEV细分的比较研究和优化,用于实时自动驾驶.

Woomin Jun1, Sungjin Lee2

  • 1Korea Electronics Technology Institute, Seongnam 13488, Republic of Korea.

Sensors (Basel, Switzerland)
|April 12, 2025
PubMed
概括

这项研究优化了嵌入式系统的实时鸟视图 (BEV) 分段. 内部图像编码器带有提升平板拍摄实现了卓越的准确性和效率,超越了以前的方法.

科学领域:

  • 计算机视觉 计算机视觉
  • 自主驾驶系统 自主驾驶系统
  • 嵌入式系统工程 嵌入式系统工程

背景情况:

  • 实时鸟视图 (BEV) 分段对于自动驾驶安全至关重要.
  • 现有的方法在嵌入式平台上平衡精度和计算效率方面面临挑战.
  • 对资源有限的环境进行基于摄像头的BEV细分的优化是一个关键的研究领域.

研究的目的:

  • 优化基于摄像头的鸟视角 (BEV) 分段技术,用于实时嵌入式系统部署.
  • 评估和比较基于深度,基于MLP和基于变压器的BEV细分方法.
  • 通过多阶段的优化过程实现高精度和低延迟.

主要方法:

  • 在nuScenes数据集上对lift-splat-shoot,HDMapNet和BEVFormer进行数学分析和比较性能评估.
  • 三阶段优化:提高准确性 (模块选择,输入分辨率,数据增强),减少延迟,优化模型大小 (模型压缩).
  • 使用InternImage-B/T编码器,EfficientNet-B0解码器和FP16量子化进行模型压缩.

主要成果:

  • 使用InternImage-B编码器和EfficientNet-B0解码器的升降板拍摄方法在448x800输入分辨率下实现了54.9mIoU.
  • 内部图像-T变体提供了高效率 (51.7毫秒延迟,159.5 MB大小) 的53.1mIoU.
关键词:
自动驾驶自动驾驶的自动驾驶.鸟的视角 鸟的视角细分化 细分化的细分化

更多相关视频

Driving Simulation in the Clinic: Testing Visual Exploratory Behavior in Daily Life Activities in Patients with Visual Field Defects
11:12

Driving Simulation in the Clinic: Testing Visual Exploratory Behavior in Daily Life Activities in Patients with Visual Field Defects

Published on: September 18, 2012

17.3K
Robotized Testing of Camera Positions to Determine Ideal Configuration for Stereo 3D Visualization of Open-Heart Surgery
05:12

Robotized Testing of Camera Positions to Determine Ideal Configuration for Stereo 3D Visualization of Open-Heart Surgery

Published on: August 12, 2021

1.9K

相关实验视频

Last Updated: May 14, 2025

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
Driving Simulation in the Clinic: Testing Visual Exploratory Behavior in Daily Life Activities in Patients with Visual Field Defects
11:12

Driving Simulation in the Clinic: Testing Visual Exploratory Behavior in Daily Life Activities in Patients with Visual Field Defects

Published on: September 18, 2012

17.3K
Robotized Testing of Camera Positions to Determine Ideal Configuration for Stereo 3D Visualization of Open-Heart Surgery
05:12

Robotized Testing of Camera Positions to Determine Ideal Configuration for Stereo 3D Visualization of Open-Heart Surgery

Published on: August 12, 2021

1.9K
  • FP16量子化减少了50%的记忆和延迟,保持了mLoU. 最优化的方法提高了mIoU的29.2%,同时减少了内存大小.

    • 结论:

    • 基于InternImage编码器的提升平板拍摄技术为嵌入式BEV细分提供了精度,延迟和模型大小之间的最佳权衡.
    • 最佳的输入分辨率取决于模型,需要仔细调整以获得最大的准确性.
    • 像FP16量子化这样的模型压缩技术对于在功率受限的嵌入式设备上部署精确的BEV细分是有效的.