一个基于视觉变压器的3D U-Net用于雷达语义细分.
1College of Marine Electrical Engineering, Dalian Maritime University, Dalian 116026, China.
Sensors (Basel, Switzerland)
|December 23, 2023
概括
这项研究引入了一种新的雷达语义细分网络,使用高维雷达热图进行自动目标识别. 这种方法比传统的点云方法提高了效率和准确性.
科学领域:
- 工程 工程师 工程师 工程师
- 计算机科学 计算机科学
- 信号处理 信号处理
背景情况:
- 目前的雷达目标识别通常使用点云数据,由于计算限制,其信息有限.
- 高维雷达数据提供了更丰富的信息,但高效地处理它仍然是一个挑战.
研究的目的:
- 提出一个语义细分网络,用于处理高维雷达数据,用于自动识别目标.
- 与现有方法相比,提高雷达数据利用的效率和准确性.
主要方法:
- 开发了一个使用高维雷达热图而不是点云数据的语义细分网络.
- 引入了一个基于视觉变压器的维度缩放模块,用于高维数据中有效的特征提取.
主要成果:
- 拟议的网络表现出卓越的性能,并且比现有的细分网络需要更少的参数.
- 在真实雷达数据集上的验证证实了雷达热图和视觉变压器方法的有效性.
结论:
- 新型雷达语义细分网络有效处理高维数据,以改进自动目标识别.
- 维度崩模块是处理高维数据转换的网络的多功能组件.
相关概念视频
01:24Vision
53.4K
Vision is the result of light being detected and transduced into neural signals by the retina of the eye. This information is then further analyzed and interpreted by the brain. First, light enters the front of the eye and is focused by the cornea and lens onto the retina—a thin sheet of neural tissue lining the back of the eye. Because of refraction through the convex lens of the eye, images are projected onto the retina upside-down and reversed.
53.4K
01:15Depth Perception and Spatial Vision
668
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.
668