多源时间深度融合为强大的端到端视觉测距
Sihang Zhang1, Congqi Cao2, Qiang Gao3
1School of Artificial Intelligence, Optics and Electronics (iOPEN), Northwestern Polytechnical University, Xi'an, PR China; School of Computer Science, Northwestern Polytechnical University, Xi'an, PR China.
这项研究引入了一种新的端到端多源视觉测距 (MVO) 模型. 它通过整合时间数据和深度信息来增强姿势估计,提高视觉测距任务的准确性和效率.
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
- 深度学习 (Deep Learning) 是一种深度学习.
背景情况:
- 端到端视觉测距 (VO) 模型提供高定位精度并减少故障.
- 当前的模型在使用全时间序列数据进行姿势优化时遇到困难.
- 现有的方法不充分地使用深度预测来应对规模约束.
研究的目的:
- 提出一个端到端的多源视觉测距 (MVO) 模型.
- 将混合VO组件动态集成到一个统一的深度学习框架中.
- 通过利用时间和深度信息来改善姿势估计.
主要方法:
- 开发了TimePoseNet,以捕捉跨序列的时间依赖性,用于时间对位映射.
- 采用波形卷积注意力机制来提取和嵌入全球深度信息.
- 在姿势估计后处理阶段共同结合的时间和深度线索.
主要成果:
- 在KITTI基准上取得了最先进的表现.
- 在UAV-2025数据集上表现出最佳性能.
- 在推理过程中保持计算效率.
结论:
- 拟议的MVO模型有效地利用时间和深度数据进行增强的姿势估计.
- 该框架提供了一种统一且可学习的视觉测距方法.
- 该方法在视觉测距精度和效率方面取得了显著的进步.
更多相关视频
07:14Tracking Infiltration Front Depth Using Time-lapse Multi-offset Gathers Collected with Array Antenna Ground Penetrating Radar
Published on: May 1, 2018
10:13Multi-Photon Time Lapse Imaging to Visualize Development in Real-time: Visualization of Migrating Neural Crest Cells in Zebrafish Embryos
Published on: August 9, 2017
相关概念视频
Nuclear Fusion
A helium nucleus has a mass that is 0.7% less than that of four hydrogen nuclei; this lost mass is converted into energy during the fusion. This reaction produces about...
Uniform Depth Channel Flow
Depth Perception and Spatial Vision
Uniform Depth Channel Flow: Problem Solving
Assessment of Ventilation II: Respiratory Depth and Rhythm
Respiratory depth measures the volume of air inhaled or exhaled during a breath. It can vary from shallow to deep and typically remains consistent when a person is at rest or asleep. Occasionally, individuals will automatically inhale deeply, known as sighing, which inflates the lungs with more air than normal breathing.
To assess respiratory depth, observe the degree of chest excursion or movement:
Tagging and Fusion Proteins
