Synthetic Highway Driving Scenarios Dataset

This dataset provides synthetic, scenario-labeled highway-driving samples generated using CARLA and parametrized by real-world statistics from a portion the MOOVE dataset. Each instance is delivered as a sequence of high-resolution frames and belongs to one of nine highway scenario classes, namely Cut-in, Cut-out, EV Lane Change, Approaching LV, EV pulling away from LV, Following LV, Following distant LV, Free ride, and Brake. The dataset supports research in driving-scenario recognition, automated-driving perception, and simulation-based pre-training.

Project page at elios-lab.github.io/synthetic_highway_scenarios/

本数据集提供经场景标注的合成高速公路驾驶样本，此类样本通过CARLA生成，并基于MOOVE数据集的部分真实世界统计数据完成参数化配置。
每个样本实例以高分辨率帧序列的形式交付，且隶属于9类高速公路场景类别之一，分别为车道切入、车道切出、电动车辆（Electric Vehicle，EV）车道变更、接近前导车辆（Leading Vehicle，LV）、电动车辆从前导车辆驶离、跟随前导车辆、远距离跟随前导车辆、自由巡航以及制动减速。
本数据集可支撑驾驶场景识别、自动驾驶感知以及基于仿真的预训练相关研究。
项目主页：elios-lab.github.io/synthetic_highway_scenarios/