SEHN (Synthetic Event-based Highway Nighttime)
收藏资源简介:
SEHN是一个基于CARLA仿真平台构建的大规模合成多模态数据集,专门针对高速公路夜间感知场景。该数据集包含多样化的环境条件,如白天、夜晚以及无人工照明的极端黑暗环境,并提供了同步的RGB图像与事件流数据,以支持多模态融合算法的研究与评估。数据集的创建过程通过高度可配置的仿真流程实现,能够精确控制交通密度和车速分布,模拟真实高速公路的复杂动态。该数据集旨在解决现有公开数据集中缺乏极端低光照条件下高速公路静态监控数据的空白,为智能交通系统中的车辆跟踪与感知任务提供关键的数据支撑,特别是在RGB传感器性能严重退化时,利用事件相机的高时间分辨率特性来提升系统的鲁棒性与可靠性。
SEHN is a large-scale synthetic multimodal dataset built on the CARLA simulation platform, specifically targeting highway nighttime perception scenarios. This dataset covers diverse environmental conditions including daytime, nighttime, and extremely dark environments without artificial lighting, and provides synchronized RGB images and event stream data to support the research and evaluation of multimodal fusion algorithms. The dataset is developed via a highly configurable simulation pipeline, which can precisely control traffic density and vehicle speed distribution to simulate the complex dynamics of real highways. This dataset aims to fill the gap that existing public datasets lack static monitoring data of highways under extremely low-light conditions, providing critical data support for vehicle tracking and perception tasks in intelligent transportation systems. Especially when the performance of RGB sensors deteriorates severely, it leverages the high temporal resolution characteristics of event cameras to enhance the robustness and reliability of the system.
由于该数据集详情页面(https://github.com/haidongwang96/SEHN)的README文件内容仅为“coming soon”,目前没有提供任何关于数据集的具体信息,因此无法提炼出有效的关键内容。

- 1Event-RGB Adaptive Tracking for Nighttime Highway Perception中山大学·智能系统工程系; 哈尔滨工业大学·智能科学与工程学院 · 2026年




