海底隧道无人化巡检病害数据集

该数据集包含使用无人化巡检设备在青岛地铁8号线采集的高清图像与三维点云数据，以及为深度学习模型训练而生成的YOLO格式病害标注文件。数据采集前通过标定板/裂缝卡进行精度标定，采集后利用PCL/OpenCV库进行去噪与去畸变处理，并由专业人员使用Labelme软件完成精细化标注，为病害自动识别算法开发提供了高质量训练素材。该数据集通过多次现场采集建立了海底隧道常见病害特征库，涵盖了不同程度的裂缝、渗漏水、锈蚀、剥落等典型病害，为构建高精度的病害智能识别系统奠定了数据基础。

This dataset contains high-definition images and 3D point cloud data collected via unmanned inspection equipment on Qingdao Metro Line 8, as well as YOLO-format defect annotation files generated for deep learning model training. Prior to data acquisition, accuracy calibration was conducted using calibration boards and crack cards. After data collection, denoising and distortion correction were performed with the PCL and OpenCV libraries, and fine-grained annotation was completed by professionals using LabelMe software, providing high-quality training materials for the development of automated defect recognition algorithms. This dataset established a feature database of common defects in undersea tunnels through multiple on-site collection campaigns, covering typical defects including cracks, water seepage and leakage, rust, spalling and others with varying degrees of severity, laying a solid data foundation for the construction of high-precision intelligent defect recognition systems.

搜集汇总

背景与挑战

背景概述

该数据集基于青岛地铁8号线的无人化巡检设备采集，包含高清图像、三维点云数据及YOLO格式的病害标注文件，经过精度标定和去噪处理。它涵盖了裂缝、渗漏水、锈蚀等典型病害，为深度学习模型训练和病害智能识别系统开发提供了高质量素材。