IDD-X

该数据集名为IDD-X，是一个大规模的双视角驾驶视频数据集，它为多个关键道路物体提供了以自我为中心的标注，旨在理解复杂交通条件下的驾驶行为。该数据集涵盖了10种不同的道路物体类别和19种解释类别，这些数据是在特定的相机分辨率和帧率下捕获的。规模上，数据集包含了85小时的双视角驾驶视频，697,000个边界框，9,000个重要物体轨迹，以及3,635个驾驶行为场景。任务的目的是进行以自我为中心的重要物体定位及其解释。

This dataset, named IDD-X, is a large-scale dual-view driving video dataset that provides ego-centric annotations for multiple critical road objects, aiming to understand driving behaviors under complex traffic conditions. It covers 10 distinct road object categories and 19 interpretation categories, with all data captured under specific camera resolutions and frame rates. In terms of scale, the dataset contains 85 hours of dual-view driving videos, 697,000 bounding boxes, 9,000 critical object trajectories, and 3,635 driving behavior scenarios. The core task of this dataset is to perform ego-centric critical object localization and its corresponding interpretation.