DADA-2000
收藏资源简介:
DADA-2000是由长安大学电子与控制工程学院和西安交通大学人工智能与机器人研究所共同创建的大型视频基准数据集,专注于驾驶事故预测。该数据集包含2000个视频片段,总计约658,476帧,覆盖54种不同类型的交通事故。数据来源广泛,包括高速公路、城市、乡村和隧道等不同场合,以及晴天、雨天和雪天等天气条件。创建过程中,首先收集了约300万帧的众包视频,经过筛选和分类后,选出658,476帧进行详细标注。DADA-2000的应用领域主要集中在驾驶安全领域,旨在通过分析驾驶员的注意力来预测和避免交通事故。
DADA-2000 is a large-scale video benchmark dataset focused on driving accident prediction, jointly created by the School of Electronic and Control Engineering of Chang'an University and the Institute of Artificial Intelligence and Robotics of Xi'an Jiaotong University. This dataset contains 2000 video clips, totaling approximately 658,476 frames, covering 54 distinct types of traffic accidents. The data is sourced from a wide range of scenarios including highways, urban areas, rural regions, tunnels and other occasions, as well as various weather conditions such as sunny, rainy and snowy days. During its development, approximately 3 million frames of crowdsourced videos were first collected, and after screening and classification, 658,476 frames were selected for detailed annotation. The main application areas of DADA-2000 are concentrated in the field of driving safety, aiming to predict and avoid traffic accidents by analyzing drivers' attention.




