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

Depth Perception and Spatial Vision01:15

Depth Perception and Spatial Vision

625
Depth perception is the ability to perceive objects three-dimensionally. It relies on two types of cues: binocular and monocular. Binocular cues depend on the combination of images from both eyes and how the eyes work together. Since the eyes are in slightly different positions, each eye captures a slightly different image. This disparity between images, known as binocular disparity, helps the brain interpret depth. When the brain compares these images, it determines the distance to an object.
625

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

Elucidation of Enhanced Lysine Sensing via Pt-Functionalized 2D WS<sub>2</sub> for Biosensing Applications.

ACS applied materials & interfaces·2026
Same author

Diffusion-based skin disease data augmentation with fine-grained detail preservation and interpolation for data diversity.

PloS one·2025
Same author

A Comparative Study and Optimization of Camera-Based BEV Segmentation for Real-Time Autonomous Driving.

Sensors (Basel, Switzerland)·2025
Same author

Selective denoising autoencoder for classification of noisy gas mixtures using 2D transition metal dichalcogenides.

Talanta·2024
Same author

Optimal Configuration of Multi-Task Learning for Autonomous Driving.

Sensors (Basel, Switzerland)·2023
Same author

Numerical Approach to Facial Palsy Using a Novel Registration Method with 3D Facial Landmark.

Sensors (Basel, Switzerland)·2022

相关实验视频

Updated: Jun 21, 2025

High-resolution, High-speed, Three-dimensional Video Imaging with Digital Fringe Projection Techniques
11:34

High-resolution, High-speed, Three-dimensional Video Imaging with Digital Fringe Projection Techniques

Published on: December 3, 2013

15.6K

合成数据增强和单眼深度估计的网络压缩技术,用于实时自动驾驶系统.

Woomin Jun1,2, Jisang Yoo2,3, Sungjin Lee1,2

  • 1Electronic Engineering, Dong Seoul University, Seongnam 13117, Republic of Korea.

Sensors (Basel, Switzerland)
|July 13, 2024
PubMed
概括

这项研究提高了单眼深度估计 (MDE) 用于自动驾驶使用新型数据增强和RMS算法. 这些方法提高了实时应用的3D感知精度和效率.

关键词:
绝对的相对误差 绝对的相对误差自动驾驶自动驾驶的自动驾驶.数据增强数据增强单眼的深度估计估计.修剪 修剪 修剪 修剪定量化定量化是什么

更多相关视频

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

2.0K
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

515

相关实验视频

Last Updated: Jun 21, 2025

High-resolution, High-speed, Three-dimensional Video Imaging with Digital Fringe Projection Techniques
11:34

High-resolution, High-speed, Three-dimensional Video Imaging with Digital Fringe Projection Techniques

Published on: December 3, 2013

15.6K
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

2.0K
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

515

科学领域:

  • 计算机视觉 计算机视觉
  • 人工智能的人工智能
  • 机器人技术 机器人技术 机器人技术

背景情况:

  • 自动驾驶的安全依赖于精确的3D图像识别.
  • 基于摄像头的深度估计正在取代LIDAR,因为检测小,遥远的物体的成本和局限性.
  • 单眼深度估计 (MDE) 为3D环境数据采集提供了具有成本效益的解决方案.

研究的目的:

  • 通过使用新的基于合成的数据增强技术,提高单眼深度估计 (MDE) 的准确性.
  • 引入实时单眼深度估计配置,考虑分辨率,效率和延迟 (RMS) 算法,以优化神经网络.
  • 为了验证在设备上的自动驾驶平台上提出的方法的性能.

主要方法:

  • 拟议的基于合成的数据增强策略:Mask,Mask-Scale和CutFlip.
  • 开发了RMS算法:一个三步过程,涉及模型选择,精度精细化与数据增强和损失函数,和网络压缩.
  • 利用量子化,修剪和FP16等技术进行模型压缩.

主要成果:

  • 合成数据增强使MDE模型的准确性提高了4.0%.
  • 在RMS约束下,IEBins模型实现了最好的REL性能 (0.0480) .
  • 数据增强 (Flip, Mask, CutFlip) 和SigLoss的组合产生了最好的REL性能 (0.0461).
  • FP16压缩将模型大小减少了83.4%,对性能和延迟的影响最小.

结论:

  • 新的数据增强和RMS算法显著提高了MDE准确性和自动驾驶的效率.
  • 拟议的方法可以在边缘设备上实现经济高效的实时3D感知.
  • 针对NVIDIA Jetson AGX Orin平台上的各种自动驾驶场景,优化了部署策略.