INDRA

INDRA数据集是由G H Patel College of Engineering and Technology创建的，专注于印度道路行人视角的交通安全问题。该数据集包含104个视频，总计超过26000帧，每帧都标注了是否安全用于道路穿越。数据集内容丰富，涵盖了多种交通场景，通过头戴式摄像机采集，确保了数据的真实性和实用性。创建过程中，研究人员手动标注了安全穿越的时间点，并提供了车辆边界框作为辅助数据。INDRA数据集主要应用于帮助视觉障碍者安全穿越印度道路，解决无交通灯或行人过街设施的地区的交通安全问题。

The INDRA dataset was developed by G H Patel College of Engineering and Technology, focusing on traffic safety issues from the perspective of road pedestrians in India. This dataset comprises 104 videos with a total of over 26,000 frames, where each frame is annotated with whether it is suitable for safe road crossing. The dataset covers diverse traffic scenarios and was collected using head-mounted cameras, ensuring the authenticity and practicality of the collected data. During its development, researchers manually annotated the time points for safe road crossing, and provided vehicle bounding boxes as auxiliary data. The INDRA dataset is primarily intended to help visually impaired people safely cross roads in India, and to address traffic safety problems in regions without traffic lights or pedestrian crossing facilities.

搜集汇总

背景与挑战

背景概述

INDRA数据集是一个专注于印度道路行人视角交通安全的数据集，包含104个视频和超过26000帧图像，每帧都标注了是否安全用于道路穿越，并通过头戴式摄像机采集以确保真实场景覆盖。该数据集主要用于帮助视觉障碍者安全穿越无交通灯或行人过街设施的印度道路，提供了手动标注的安全时间点和车辆边界框作为辅助数据。