建築的抗震能力一直以來都是土木工程領域的重要研究議題，對於依據舊版建築設計規範興建的老舊建築而言，地震帶來的結構損壞更是不容忽視，為提升災後初步評估之效率與準確性，本研究將建立一套基於深度學習的建築損壞等級分類模型，以降低人力及時間成本，並將更多資源投入在災後救援與重建工作中。
真實世界的人工標註資料集非常缺乏，因此採用合成資料是個可行的替代方案，本研究使用基於離散元素法的Blender物理引擎結合Bullet Constraint Builder 模組來模擬地震對建築造成的不同程度損壞。透過輸入各種人工地震參數，生成多組具代表性的建築損壞模型，依據其損壞程度進行人工標註分類，之後將其轉換為點雲格式並建立損壞建築資料集。接著，本研究使用PointNet深度學習模型來進行訓練與驗證，評估其於地震後建築損壞分類任務中的效能和可能性。

The seismic damage of buildings has been a critical concern for humanity. For older buildings constructed under outdated design codes, the potential structural damage caused by earthquakes is especially significant. To enhance the efficiency of post-earthquake assessments, this study would like to develop a point-cloud-based deep learning model for automation of classification for evaluating the damage levels of buildings. This model may reduce labor and time costs, enabling more resources to invest in search and rescue.
Due to the scarcity of human-annotated data, this research uses synthetic data as the training dataset for the deep learning model. Blender, along with the Bullet Constraint Builder (BCB), which is based on the Discrete Element Method (DEM), is used to simulate various levels of seismic damage on building models. By varying a set of artificial ground‐motion parameters, we generate many damaged building models, each of which will be converted into a point cloud after manually annotation according to its damage level. This labeled dataset then serves as the basis for training the PointNet framework to classify post‐earthquake damage levels. Model performance is evaluated against realistic scenarios to assess both its predictive accuracy and practical applicability.
The research workflow encompasses the collection of structural parameters from aging buildings, manual annotation of damage levels, construction of damage-level-specific datasets, and modification of open-source code to accommodate the proposed methodology.

[1] L. Manchun , C. Liang, Gong.JianYa, L. Yongxue, C, Zhenjie, L. Feixue, C.Gang, C. Dong, Song. Xiaogang. " Post-earthquake assessment of building damage degree using LiDAR data and imagery". Science in China Series E： Technological Sciences. 51. 133-143. 10.1007/s11431-008-6014-1,2009.
[2] D. P. Kingma and J. Ba, “Adam: A Method for Stochastic Optimization,” arXiv:1412.6980, 2014.
[3] Z. Lu, X. He, and Y. Zhou, "Discrete element method-based collapse simulation, validation and application to frame structures," Structure and Infrastructure Engineering, vol. 14, no. 5, pp. 538-549, 2017.
[4] FEMA, "Post-disaster Building Safety Evaluation Guidance – Report on the Current State of Practice, including Recommendations Related to Structural and Nonstructural Safety and Habitability," FEMA -2055, 2019.
[5] A. B. Khajwalcheng and N. Arash, “Multi-view deep learning for reliable post-disaster damage classification,” arXiv:2201.12345, Jan. 2022.
[6] Community, B. O. "Blender – a 3D modeling and rendering package, "Stichting Blender Foundation. Blender 2.79b User Manual, 2018.
[7] J. Xu, P. Khaitan, "Machine learning-based damage assessment for disaster relief," Google Research, 2020.
[8] M. Díaz-Medina, J. M. Fuertes, R. J. Segura-Sánchez, M. Lucena, and C. J. Ogayar-Anguita, "LiDAR attribute based point cloud labeling using CNNs with 3D convolution layers," Computers & Geosciences, vol. 180, p. 105453, 2023.
[9] T. Hackel, J. D. Wegner, K. Schindler, "Semantic3D.net： A new large-scale point cloud classification benchmark, " ETH Zurich, 2017.
[10] UN OCHA. "Bullet Constraints Builder (BCB) for Blender." GitHub Repository, 2018.
[11] R. Q. Charles, H. Su, M. Kaichun and L. J. Guibas, "PointNet： Deep Learning on Point Sets for 3D Classification and Segmentation," 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Honolulu, HI, USA, 2017, pp. 77-85, doi： 10.1109/CVPR.2017.16.
[12] D. Maturana and S. Scherer, "VoxNet： A 3D Convolutional Neural Network for real-time object recognition," 2015 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Hamburg, Germany, 2015, pp. 922-928, doi： 10.1109/IROS.2015.7353481.
[13] Z. Wu, S. Song, A. Khosla, F. Yu, L. Zhang, X. Tang, and J. Xiao, " 3D ShapeNets： A deep representation for volumetric shapes, " Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2015.
[14] H. Su, S. Maji, E. Kalogerakis, and E. G. Learned-Miller. " Multi-view convolutional neural networks for 3d shape recognition". In Proc. ICCV, to appear, 2015.
[15] C. R. Qi, L. Yi, H. Su, and L. J. Guibas, “PointNet++: Deep hierarchical feature learning on point sets in a metric space,” arXiv:1706.02413, Jun. 2017.
[16] H. Xiu, T. Shinohara, M. Matsuoka, M. Inoguchi, K. Kawabe, K. Horie. "Collapsed Building Detection Using 3D Point Clouds and Deep Learning". Remote Sensing. 2020
[18] Y. Eldar, M. Lindenbaum, M. Porat and Y. Y. Zeevi, "The Farthest Point Strategy for Progressive Image Sampling," Proceedings of the 12th IAPR International Conference on Pattern Recognition, Vol. 2 - Conference B： Computer Vision & Image Processing. (Cat. No.94CH3440-5), Jerusalem, Israel, 1994, pp. 93-97 vol.3, doi： 10.1109/ICPR.1994.577129.
[19] A. Bagheri, M. Rezaei, M. Noori, A. M. Nezhad, and M. Hammad, "As-inspected modeling of concrete surface defects from raw point clouds using deep learning," Automation in Construction, vol. 157, 2024.
[20] Y. Wang, Y. Sun, Z. Liu, S. E. Sarma, M. M. Bronstein, and J. M. Solomon, "Dynamic Graph CNN for learning on point clouds," arXiv, 2019.
[21] L. Winiwarter, A. M. Esmorís Pena, H. Weiser, K. Anders, J. Martínez Sánchez, M. Searle, and B. Höfle, “Virtual laser scanning with HELIOS++: A novel take on ray tracing–based simulation of topographic full-waveform 3D laser scanning,” Remote Sensing of Environment, vol. 269, p. 112772, 2022.
[22] 鄭亦翔，「中高層建築倒塌資料庫模擬與驗證」，國立成功大學土木工程研究所碩士論文，台南市，2022年。
[23] 內政部國土管理署，「地震災後建築物緊急評估明細表」,內政部國土管理署，台北市，2023年版。
[24] 林育正，「點雲建築資訊模型倒塌模式生成與預測」，國立成功大學土木工程研究所碩士論文，台南市，2018年。
[25] 吳威霖，「以點雲資料進行倒塌建物構件模組之快速重建」，國立成功大學土木工程研究所碩士論文，台南市，2022年。
[26] 黃士懷，「中高層建築混合倒塌與存活空間模擬」，國立成功大學土木工程研究所碩士論文，台南市，2024年。
[27] 財團法人住宅地震保險基金，「住宅建築物地震毀損程度評估方法及判定準則」，財團法人住宅地震保險基金，2022年版。