Monitoring Roofs of Traditional Buildings using Remote Sensing

气候变化对英国和世界的传统建筑造成了巨大的破坏。为了确保建筑物得到适当的维护和居住者的安全停留，有必要对维护解决方案进行监测和分析。传统的屋顶状况监测和分析依靠人工操作，费时、费力且不安全(Curtis&Kennedy，2016)。为了克服这些限制，开发了遥感等现代监测方法，这类扫描到BIM和扫描VS-BIM应用包括对建筑物、桥梁、管道、隧道进行状况评估(Koch等人，2015)，监测建筑物沉降(Dai&Lu，2010)，以及检测建筑物屋顶损坏(Vetrivel等人，2018)。深度学习等高级分析方法也被应用于土木工程中，例如混凝土结构表面的裂缝检测(Yokyama等人，2017年)。然而，目前还没有一个完整的库包含不同类型和不同劣化程度的传统建筑屋顶数据，供研究和工业使用。此外，现有的土木工程深度学习研究大多只关注裂缝等对象的识别，而从建筑全生命周期分析的角度提出维修解决方案的较少。

Climate change has caused great damages in traditional buildings in the UK and the world. To ensure that the buildings are maintained properly and safe for occupants to stay, it is necessary to perform monitoring and analyzing maintenance solutions. Traditional roof condition monitoring and analysis rely on manual operation, which is time-consuming, labor-intensive, and unsafe (Curtis & Kennedy, 2016). To overcome these limitations, modern monitoring methods like remote sensing have been developed, such Scan-to-BIM and Scan-vs-BIM applications include condition assessment on buildings, bridges, pipes, tunnels (Koch et al., 2015), monitoring of building settlement (Dai & Lu, 2010) and detecting building roof damage (Vetrivel et al., 2018). Advanced analysis methods like deep learning have also been applied in civil engineering, such as crack detection on concrete structure surface (Yokoyama et al., 2017). However, there is not a complete library containing traditional building roof data of different types and with different deteriorations for research and industry use. Besides, most existing studies of civil engineering on deep learning only focus on recognizing objects like cracks, but few of them are from the perspective of building life-cycle analysis like raising maintenance solutions.