一个合成的数字城市数据集,用于深度估计模型的稳定性和概括性
Jihao Li1, Jincheng Hu1, Yanjun Huang2
1Department of Aeronautical and Automotive Engineering, Loughborough University, Leicestershire, LE11 3TU, UK.
Scientific data
|March 17, 2024
概括
一个新的合成数字城市数据集 (SDCD) 提供了高分辨率图像和深度地图,用于训练自动驾驶模型. 在SDCD上训练的模型在各种不利条件下改善了远程深度估计.
科学领域:
- 计算机视觉 计算机视觉
- 机器学习 机器学习
- 自主驾驶系统 自主驾驶系统
背景情况:
- 现有的单眼深度估计数据集缺乏图像数量和多样性.
- 当前的数据集通常具有低分辨率图像和稀疏的深度图,限制了模型性能.
- 当前数据集中的驾驶条件不能充分反映现实世界的复杂性.
研究的目的:
- 为单眼深度估计引入一个全面的合成数字城市数据集 (SDCD).
- 提高自动驾驶深度估计模型的准确性和稳定性.
- 为评估各种不利驾驶场景下的模型性能提供一个基准.
主要方法:
- 创建了一个大规模的数据集 (930K图像),具有高分辨率的RGB图像和完美的深度图.
- 模拟了6种不同的天气条件和6种不利的数据传输干扰.
- 使用拟议的SDCD训练和评估深度估计模型,并与KITTI等现有数据集进行比较.
主要成果:
- 与在KITTI上训练的模型相比,在SDCD上训练的模型表现出更高的远程深度估计准确度,清晰度和流性.
- 建立了一个基准来分析不同不利驾驶条件下的模型性能.
- 合成方法使得在严峻的条件下能够生成数据,并且具有完美的基本真实性.
结论:
- 合成数字城市数据集 (SDCD) 有效地解决了单眼深度估计现有数据集的局限性.
- SDCD显著提高了自动驾驶系统的深度估计能力,特别是在具有挑战性的环境中.
- 生成的数据集为推进强大的自主感知研究提供了宝贵的资源.
相关概念视频
Depth Perception and Spatial Vision
646
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.
646
Methods of Obtaining Topography
65
Topography involves measuring and mapping land elevations, natural features, and artificial structures to create accurate representations of the terrain. Topographic surveying relies on traditional and modern methods, each with distinct advantages and limitations.Traditional Surveying Methods:Transit stadia surveys and plane table surveys were widely used traditional surveying methods. These techniques relied on instruments like theodolites and stadia rods for measuring distances and angles,...
65
Distance Measurements by Taping
37
Tapes are essential in surveying for accurate, durable, and short-distance measurements. Made from lightweight, nylon-coated steel, they offer flexibility and strength for rugged outdoor use. The nylon coating protects against rust and wear, extending the tape's life. Standard lengths, around 30 meters, are marked in meters and millimeters for precision.Surveyors select tapes based on site conditions and accuracy needs. Lightweight, nylon-coated tapes are commonly used for ease of handling and...
37
Estimation of the Physical Quantities
4.3K
On many occasions, physicists, other scientists, and engineers need to make estimates of a particular quantity. These are sometimes referred to as guesstimates, order-of-magnitude approximations, back-of-the-envelope calculations, or Fermi calculations. The physicist Enrico Fermi was famous for his ability to estimate various kinds of data with surprising precision. Estimating does not mean guessing a number or a formula at random. Instead, estimation means using prior experience and sound...
4.3K
Modeling and Similitude
266
Scaled modeling is a fundamental technique in engineering, enabling the study of large and complex systems by creating smaller, manageable replicas that recreate critical characteristics of the original. In hydrology and civil infrastructure, for example, scaled models of dams help analyze water flow, turbulence, and pressure. This method allows for accurate predictions of real-world behavior within a controlled environment, significantly reducing the cost and time involved in full-scale...
266
Survival Tree
84
Survival trees are a non-parametric method used in survival analysis to model the relationship between a set of covariates and the time until an event of interest occurs, often referred to as the "time-to-event" or "survival time." This method is particularly useful when dealing with censored data, where the event has not occurred for some individuals by the end of the study period, or when the exact time of the event is unknown.
Building a Survival Tree
Constructing a...
Building a Survival Tree
Constructing a...
84


