轨道地面固定站点红外图像数据

本数据集是雄安R1线一段铁路区域的行人入侵检测红外图像数据集，红外图像由IR-Pilot640X红外相机所采集，数据集包括3.3GB的图像数据文件，涵盖了不同天气条件下，不同距离下的行人入侵实际图片，并经过细致清洗，确保了数据集图片之间的差异性与多样性。该数据集为行人入侵轨道检测领域提供了宝贵的红外资源，对于推动铁路周界行人入侵检测等研究任务具有重大潜力。通过公开该数据集，有助于基于红外图像的行人检测算法的研究发展，为科研人员提供宝贵的数据资源。在雄安R1线实验，在地面固定站点安装IR-Pilot640X红外相机。传感器的频率均为10HZ，所有传感器通过GNSSPPS信号与硬件触发模块实现微秒级时间同步，空间坐标经外参标定后统一至本地东北天（ENU）坐标系。红外图像数据由运行于 Ubuntu20.04+ROSNoetic 的主控计算机采集，主控机器为爱视图灵CES-RUG-ORIN-701，使用 usb_cam  驱动节点发布话题，并通过rosbagrecord 命令将话题同步入 .bag 文件。针对雄安新区城市轨道交通规划和线路环境特点，开展多源融合的路网净空安全智能感知体系与装备研究，实现对线路障碍物、周界入侵、自然灾害等威胁行车安全事件的全面感知，为保障雄安新区轨道交通的安全运营提供支撑。数据量3.3G。

This dataset is an infrared image dataset for pedestrian intrusion detection in a section of Xiong'an Rail Transit Line R1. Infrared images were acquired using the IR-Pilot640X infrared camera. The dataset encompasses 3.3 GB of image data, including actual pedestrian intrusion images captured under diverse weather conditions and at varying distances. It has undergone meticulous cleaning to guarantee the diversity and uniqueness of the included images. This dataset offers valuable infrared resources for the field of railway pedestrian intrusion detection, and exhibits significant potential for advancing research initiatives such as railway perimeter pedestrian intrusion detection.

Public release of this dataset will facilitate the research and development of infrared image-based pedestrian detection algorithms, providing valuable data resources for researchers.

For the field experiment on Xiong'an Rail Transit Line R1, IR-Pilot640X infrared cameras were installed at fixed ground-based stations. All sensors operate at a sampling frequency of 10 Hz, and are synchronized at the microsecond level via the GNSSPPS signal and hardware trigger module. Spatial coordinates are unified to the local East-North-Up (ENU) coordinate system following extrinsic calibration. The infrared image data was collected by a master control computer running Ubuntu 20.04 and ROS Noetic. The master control unit is the Aiviewling CES-RUG-ORIN-701, which uses the usb_cam driver node to publish ROS topics, and synchronizes these topics into .bag files via the rosbag record command.

In response to the characteristics of urban rail transit planning and line environments in Xiong'an New Area, research on multi-source fusion-based intelligent perception systems and equipment for network clearance safety is being conducted. This research aims to achieve comprehensive perception of traffic safety threats including line obstacles, perimeter intrusions, natural disasters and other hazardous events, providing support for ensuring the safe operation of rail transit in Xiong'an New Area. The total size of the dataset is 3.3 GB.

搜集汇总

背景与挑战

背景概述

该数据集是雄安R1线铁路区域的行人入侵检测红外图像数据集，由IR-Pilot640X红外相机采集，包含3.3GB图像数据，覆盖不同天气和距离条件，并经过清洗以确保多样性。数据集旨在为铁路周界行人入侵检测研究提供红外图像资源，支持智能感知算法开发，数据量达4.07GB，包含12070个文件，格式为jpg和docx，来源于国家重点研发计划项目。