TACK Tunnel Data (TTD)
收藏资源简介:
TACK隧道数据集(TTD)是由瑞典皇家理工学院和罗马第一大学联合创建的公开数据集,旨在支持隧道缺陷检测的深度学习研究。该数据集包含3774张高分辨率图像,涵盖三种不同类型隧道中的裂缝、渗水和淋滤等典型缺陷,每张图像分辨率达2448×2048像素。数据通过配备激光雷达和红外相机的移动测绘系统采集,并采用半自动化标注平台进行精细标注。该数据集专门用于解决隧道结构健康监测中的自动化缺陷识别难题,为监督、半监督和无监督学习方法提供了跨隧道类型泛化能力评估的基准数据。
The TACK Tunnel Dataset (TTD) is a public dataset jointly developed by KTH Royal Institute of Technology and Sapienza University of Rome, aiming to support deep learning research for tunnel defect detection. This dataset includes 3774 high-resolution images, covering typical defects such as cracks, water seepage and leaching in three distinct tunnel types, with each image having a resolution of 2448×2048 pixels. The data was collected using a mobile mapping system equipped with LiDAR and infrared cameras, and underwent fine annotation via a semi-automated annotation platform. This dataset is specifically designed to address the challenge of automated defect recognition in tunnel structural health monitoring, and provides benchmark data for evaluating the cross-tunnel-type generalization capability of supervised, semi-supervised and unsupervised learning methods.

- 1TACK Tunnel Data (TTD): A Benchmark Dataset for Deep Learning-Based Defect Detection in Tunnels瑞典皇家理工学院, 罗马第一大学 · 2025年



