Anticipation of Traffic Accident (AoTA)
收藏资源简介:
AoTA数据集是针对交通事故预测而设计的一个新的基准数据集,旨在更好地捕捉现实世界中的驾驶风险。该数据集包含大量真实世界的交通事故场景,并附带丰富的注释信息。数据集的创建过程结合了视觉语言模型和世界模型,以确保生成的场景数据与原始数据集具有相同的特征分布。数据集主要用于自动驾驶系统的交通事故预测,以提高系统的准确性和提前预警时间。
The AoTA dataset is a novel benchmark dataset designed for traffic accident prediction, aiming to better capture driving risks in real-world scenarios. It contains a large number of real-world traffic accident scenarios with rich annotated information. The creation of this dataset combines vision-language models and world models to ensure that the generated scenario data shares the same feature distribution as the original dataset. This dataset is primarily used for traffic accident prediction in autonomous driving systems, to improve the system accuracy and extend their early warning lead time.

- 1World Model-Based End-to-End Scene Generation for Accident Anticipation in Autonomous Driving澳门大学 · 2025年



