HDSVT: High-Density Semantic Vehicle Trajectory Dataset Based on a Cosmopolitan City Bridge

Trajectory prediction is crucial for autonomous driving, necessitating robust models supported by high-quality datasets. Existing datasets often lack long trajectories and high vehicle density, limiting their use in complex scenarios. We introduce the High-Density Semantic Vehicle Trajectory Dataset (HDSVT), collected via UAVs over Guangzhou Bridge. This dataset features higher vehicle density, extended trajectory lengths, and semantic enhancements, promoting in-depth vehicle trajectory analysis. It includes original UAV videos, processed data with (i) pixel coordinates of vehicle trajectory and lane lines in each video, (ii) corresponding geographic coordinates and (iii) semantic promotion for trajectories and motions. The dataset supports diverse driving behaviors and complex interactions, applicable to driving decision-making, traffic control, and long-term small object tracking.

轨迹预测对于自动驾驶而言至关重要，亟需依托高质量数据集支撑鲁棒的预测模型。现有数据集普遍存在轨迹长度不足、车辆密度偏低的缺陷，限制了其在复杂交通场景中的应用范围。本文提出高密度语义车辆轨迹数据集（High-Density Semantic Vehicle Trajectory Dataset, HDSVT），该数据集通过无人机（Unmanned Aerial Vehicle, UAV）航拍获取自广州大桥上空。该数据集具备更高的车辆密度、更长的轨迹长度以及语义增强特性，可推动车辆轨迹的深度分析研究。数据集包含原始无人机航拍视频，以及三类处理后的数据：(i) 各视频中车辆轨迹与车道线的像素坐标，(ii) 对应的地理坐标，(iii) 轨迹与运动的语义增强信息。该数据集涵盖多样化的驾驶行为与复杂的交通交互场景，可应用于驾驶决策、交通管控以及长时序小目标跟踪等研究方向。

搜集汇总

背景与挑战

背景概述

HDSVT是一个高密度语义车辆轨迹数据集，基于广州大桥的无人机采集数据，具有高车辆密度、长轨迹和语义增强特点，适用于自动驾驶和交通控制等复杂场景分析。