IDD-X

IDD-X是一个大规模的双视角驾驶视频数据集，旨在解决发展中国家复杂且非结构化的交通环境中自动驾驶系统的解释性问题。该数据集包含697,000个边界框、9,000个重要对象轨迹，覆盖10个类别和19个解释标签类别，提供了全面的自我相对标注。IDD-X不仅捕捉了前视图信息，还包含了后视图信息，以更完整地代表驾驶环境。数据集的创建涉及在印度海得拉巴城市及其周边地区的不同道路和气候条件下录制85小时的视频，使用CVAT工具进行标注。IDD-X的应用领域主要集中在研究道路条件和周围实体如何影响复杂交通情况下的驾驶行为，为自动驾驶系统的可靠性和效率提供了基础。

IDD-X is a large-scale dual-view driving video dataset aimed at addressing the interpretability issues of autonomous driving systems in complex and unstructured traffic environments of developing countries. This dataset contains 697,000 bounding boxes and 9,000 critical object trajectories, covering 10 categories and 19 explanatory label categories, and provides comprehensive ego-centric annotations. IDD-X captures not only front-view information but also rear-view information to represent the driving environment more comprehensively. The creation of the dataset involved recording 85 hours of videos under various road and climatic conditions in Hyderabad, India, and its surrounding areas, with annotations performed using the CVAT tool. The application scenarios of IDD-X mainly focus on researching how road conditions and surrounding entities affect driving behaviors in complex traffic situations, providing a foundation for improving the reliability and efficiency of autonomous driving systems.