Rukaia/TACK_Tunnel_Data
收藏资源简介:
--- language: - en pretty_name: TTD --- # TACK Tunnel Data (TTD): A Benchmark Dataset for Deep Learning-Based Defect Detection in Tunnels Tunnels are essential elements of transportation infrastructure, but are increasingly affected by ageing and deterioration mechanisms such as cracking. Regular inspections are required to ensure their safety, yet traditional manual procedures are time-consuming, subjective, and costly. Recent advances in mobile mapping systems and Deep Learning (DL) enable automated visual inspections. However, their effectiveness is limited by the scarcity of tunnel datasets. This is the official repository of a new publicly available dataset containing annotated images of three different tunnel linings, capturing typical defects: cracks, leaching, and water infiltration. The dataset is designed to support supervised, semi-supervised, and unsupervised DL methods for defect detection and segmentation. Its diversity in texture and construction techniques also enables investigation of model generalization and transferability across tunnel types. By addressing the critical lack of domain-specific data, this dataset contributes to advancing automated tunnel inspection and promoting safer, more efficient infrastructure maintenance strategies. ## Dataset Description Data are collected from three tunnels (Tunnel A, Tunnel B, Tunnel C) using a mobile mapping system equipped with high-resolution cameras. The three tunnels exhibit different surfaces due to the construction methods used, as visible in Figure 1. <p align="left"> <img src="figures/figure1.png" width="600"/> </p> <p align="left"> <em>Figure 1: Example of high-resolution images depicting the surface of the three tunnels.</em> </p> The dataset includes three classes of defects from each tunnel. In total, 785 images with cracks, 197 images with water and 316 images with leaching (Table 1). <p align="left"> <img src="figures/table1.png" width="600"/> </p> <p align="left"> <em>Table 1: Description of the complete dataset, divided into three tunnel types and three classes.</em> </p> For each image, a segmentation mask is provided. Masks were generated manually using the SuperAnnotate platform. For any additional detail and technical validation of the dataset, please refer to the preprint https://arxiv.org/abs/2512.14477. ## Citation If you use this dataset in your research, please cite: ```bibtex @misc{sjölander2025tacktunneldatattd, title={TACK Tunnel Data (TTD): A Benchmark Dataset for Deep Learning-Based Defect Detection in Tunnels}, author={Andreas Sjölander and Valeria Belloni and Robel Fekadu and Andrea Nascetti}, year={2025}, eprint={2512.14477}, archivePrefix={arXiv}, primaryClass={cs.CV}, url={https://arxiv.org/abs/2512.14477}, }
--- 语言: - 英语 友好名称:TTD --- # TACK隧道数据(TTD):面向隧道深度学习缺陷检测的基准数据集 隧道作为交通基础设施的核心组成部分,正日益受到老化及开裂等劣化机制的影响。为保障隧道安全,需定期开展巡检,但传统人工巡检流程耗时久、主观性强且成本高昂。近年来,移动测绘系统与深度学习(Deep Learning, DL)技术的进步推动了自动化视觉巡检的实现。然而,隧道领域公开数据集的匮乏限制了此类技术的应用效果。本仓库为全新公开数据集的官方托管仓库,该数据集包含三种不同隧道衬砌的标注图像,涵盖开裂、渗沥与渗水三类典型隧道缺陷。本数据集旨在支持用于缺陷检测与分割的有监督、半监督及无监督深度学习方法。其纹理与施工工艺的多样性,也为研究模型在不同隧道类型间的泛化能力与迁移性提供了支撑。针对当前领域专属数据严重不足的痛点,本数据集有助于推动自动化隧道巡检技术的发展,并助力更安全、高效的基础设施运维策略落地。 ## 数据集说明 本数据集通过搭载高分辨率相机的移动测绘系统,从隧道A、隧道B、隧道C三座隧道中采集得到。三座隧道因施工工艺不同,表面纹理存在显著差异,具体可参见图1。 <p align="left"> <img src="figures/figure1.png" width="600"/> </p> <p align="left"> <em>图1:三座隧道表面的高分辨率图像示例。</em> </p> 本数据集为每座隧道均标注了三类缺陷。总计包含裂缝图像785张、渗水图像197张以及渗沥图像316张(详见表1)。 <p align="left"> <img src="figures/table1.png" width="600"/> </p> <p align="left"> <em>表1:完整数据集的分类说明,按三类隧道与三类缺陷进行划分。</em> </p> 每幅图像均配有分割掩码(segmentation mask),掩码通过SuperAnnotate平台手动标注生成。 如需了解数据集的更多细节与技术验证内容,请参阅预印本论文:https://arxiv.org/abs/2512.14477。 ## 引用说明 如果您在研究中使用本数据集,请引用如下文献: bibtex @misc{sjölander2025tacktunneldatattd, title={TACK Tunnel Data (TTD): A Benchmark Dataset for Deep Learning-Based Defect Detection in Tunnels}, author={Andreas Sjölander and Valeria Belloni and Robel Fekadu and Andrea Nascetti}, year={2025}, eprint={2512.14477}, archivePrefix={arXiv}, primaryClass={cs.CV}, url={https://arxiv.org/abs/2512.14477}, }




