V2X交通危险事件测试数据集

V2X交通危险事件测试数据集用于评估模型对交通危险事件检测效果，目标场景包括超速、逆行、停车占道、行人闯入、非机动车闯入、机动车闯红灯、不按导向行驶、拥堵、事故、施工，数据集包括使用CARLA仿真器生成的特定异常事件数据和从真实交通数据集（如DoTA、CODA、TrafficNet）筛选的异常场景数据，所有图片和视频数据均经过标注、清洗、检测后存储，确保数据的高质量。

The V2X traffic hazard event test dataset is developed to evaluate the performance of models in traffic hazard event detection. The target scenarios include speeding, reverse driving, road occupation by parking, pedestrian intrusion, non-motor vehicle intrusion, motor vehicles running red lights, failing to drive in designated lanes, traffic congestion, traffic accidents, and construction zones. The dataset consists of two parts: specific abnormal event data generated using the CARLA simulator, and abnormal scene data screened from real-world traffic datasets such as DoTA, CODA, and TrafficNet. All image and video data have undergone annotation, cleaning and quality inspection before being stored, ensuring high data quality.

搜集汇总

背景与挑战

背景概述

V2X交通危险事件测试数据集是一个专为评估交通危险事件检测模型设计的高质量数据集，涵盖超速、逆行、停车占道等10类常见交通异常场景。数据集包含6.36GB的4777个文件，融合了CARLA仿真器生成的模拟数据和从真实交通数据集中筛选的真实场景数据，所有数据均经过严格标注和清洗，适用于V2X和自动驾驶领域的研究与测试。