Autonomous Driving Sensor Dataset

该数据集涵盖了三个月内19个测量日的数据，为自动驾驶技术提供了一个多样化且真实的测试环境。它包含了来自摄像头和雷达的传感器追踪数据以及相应的标注信息，同时覆盖了城市和高速公路的驾驶场景。为了防止数据分割之间的相关性，该数据集通过按录制日期分组和分配数据，将每个领域分为75%的训练数据、15%的验证数据和10%的测试数据，并力求在两个领域中进行训练时保持样本数量相等。总样本量为1,849,356个，其中高速公路场景有922,081个，城市场景有927,275个。该数据集的任务是针对自动驾驶中的目标检测与跟踪。

This dataset comprises data collected over 19 measurement days within a three-month timeframe, delivering a diverse and realistic testing environment for autonomous driving technologies. It contains sensor tracking data from cameras and radars alongside corresponding annotation information, and covers both urban and highway driving scenarios. To eliminate correlation among data splits, this dataset groups and allocates data by recording date to partition each domain into 75% training data, 15% validation data, and 10% test data, while aiming to maintain equal sample sizes when training across both domains. The total sample size amounts to 1,849,356, with 922,081 samples from highway scenarios and 927,275 samples from urban scenarios. The task of this dataset centers on object detection and tracking in autonomous driving.