Adver-City
收藏资源简介:
Adver-City是由皇后大学计算学院创建的第一个开源多模态协作感知数据集,专注于恶劣天气条件。该数据集包含超过24,000帧和890,000个标注,涵盖110个独特场景,涉及六种不同的天气条件。数据集内容包括来自车辆和路边单元的LiDAR、RGB和语义分割相机、GNSS和IMU数据。创建过程基于CARLA模拟器和OpenCDA框架,场景设计基于真实事故报告,旨在模拟恶劣天气和低能见度条件下的最相关道路配置。该数据集主要用于测试和改进自动驾驶车辆在恶劣天气条件下的感知模型,解决传感器性能下降和物体检测困难的问题。
Adver-City is the first open-source multimodal collaborative perception dataset developed by the School of Computing at Queen's University, with a focus on adverse weather conditions. This dataset encompasses over 24,000 frames and 890,000 annotations, covering 110 unique scenarios across six distinct weather conditions. It comprises LiDAR data, RGB camera data, semantic segmentation camera data, GNSS data, and IMU data collected from both vehicles and roadside units. Built upon the CARLA simulator and the OpenCDA framework, the dataset's scenarios are designed based on real-world accident reports, aiming to replicate the most relevant road configurations under adverse weather and low-visibility conditions. This dataset is primarily intended for testing and enhancing perception models for autonomous vehicles in adverse weather environments, addressing the challenges of degraded sensor performance and difficult object detection.
- 1Adver-City: Open-Source Multi-Modal Dataset for Collaborative Perception Under Adverse Weather Conditions皇后大学计算学院 · 2024年



