Pedestrian Patterns Dataset
收藏arXiv2020-01-07 更新2024-06-21 收录
下载链接：
https://sites.psu.edu/real/PedestrianPatternsDataset
下载链接
链接失效反馈官方服务：
资源简介：
Pedestrian Patterns Dataset是由宾夕法尼亚州立大学创建的，旨在捕捉自动驾驶路线上的行人和社会行为模式。该数据集包含超过600GB的全高清视频和GPS数据，记录了三周内在宾夕法尼亚州立学院的三条不同路线上，每天同一时间段的行人密度变化。数据集通过Fast R-CNN行人检测方法分析视频，以评估行人密度并生成行人热图。此数据集不仅用于评估自动驾驶风险，还用于长期基于视觉的移动机器人和自动驾驶车辆定位研究。
Created by The Pennsylvania State University, the Pedestrian Patterns Dataset is designed to capture pedestrian and social behavioral patterns along autonomous driving routes. The dataset contains over 600 GB of full high-definition video and GPS data, which records changes in pedestrian density at the same daily time slot across three distinct routes over a three-week period at State College, Pennsylvania. The dataset analyzes the video via the Fast R-CNN pedestrian detection method to evaluate pedestrian density and generate pedestrian heatmaps. This dataset is not only used for assessing autonomous driving risks, but also supports long-term vision-based mobile robot and autonomous vehicle localization research.
提供机构：
宾夕法尼亚州立大学
创建时间：
2020-01-07
