ROOF3D: A REAL AND SYNTHETIC DATA COLLECTION FOR INDIVIDUAL BUILDING ROOF PLANE AND BUILDING SECTIONS DETECTION

被引:1
|
作者
Schuegraf, P. [1 ]
Reyes, M. Fuentes [1 ]
Xu, Y. [1 ]
Bittner, K. [1 ]
机构
[1] German Aerosp Ctr DLR, Remote Sensing Technol Inst, Wessling, Germany
来源
GEOSPATIAL WEEK 2023, VOL. 10-1 | 2023年
关键词
Building instance segmentation; roof plane instance segmentation; building vectorization; semantic segmentation; 3D reconstruction;
D O I
10.5194/isprs-annals-X-1-W1-2023-971-2023
中图分类号
K85 [文物考古];
学科分类号
0601 ;
摘要
Deep learning is a powerful tool to extract both individual building and roof plane polygons. But deep learning requires a large amount of labeled data. Hence, publicly available level of detail (LoD)-2 datasets are a natural choice to train fully convolutional neural networks (FCNs) models for both building section and roof plane instance segmentation. Since publicly available datasets are often automatically derived, e.g. based on laser scanning, they lack on annotation accuracy. To complement such a dataset, we introduce manually annotated and synthetically generated data. Manually annotated data is domain-specific and has a high annotation quality but is expensive to obtain. Synthetically generated data has high-quality annotations by definition, but lacks domain-specificity. Moreover, we not only detect individual building section instances, but also roof plane instances. We predict separations not only between individual buildings, but also by a class that describes the line which separates roof planes. The predicted building and roof plane instances are polygonized by a simple tree search algorithm. To achieve more regular polygons, we utilize the Douglas-Peucker polygon simplification algorithm. We describe our dataset in detail to allow comparability between successive methods. To facilitate future works in building and roof plane prediction, our Roof3D dataset is accessible at https://github.com/dlrPHS/GPUB.
引用
收藏
页码:971 / 979
页数:9
相关论文
共 46 条
  • [41] Generation of 3D building model using 3D line detection scheme based on line fitting of elevation data
    Woo, DM
    Han, SS
    Jung, YK
    Lee, KW
    ADVANCES IN MULTIMEDIA INFORMATION PROCESSING - PCM 2005, PT 1, 2005, 3767 : 559 - 569
  • [42] 3D seismic fault detection using the Gaussian process regression, a study on synthetic and real 3D seismic data
    Noori, Maryam
    Hassani, Hossein
    Javaherian, Abdolrahim
    Amindavar, Hamidreza
    JOURNAL OF PETROLEUM SCIENCE AND ENGINEERING, 2020, 195
  • [43] Seismic channel edge detection using 3D shearletsa study on synthetic and real channelised 3D seismic data
    Karbalaali, Haleh
    Javaheriani, Abdolrahim
    Dahlke, Stephan
    Reisenhofer, Rafael
    Torabi, Siyavash
    GEOPHYSICAL PROSPECTING, 2018, 66 (07) : 1272 - 1289
  • [44] Real-time drill monitoring and control using building information models augmented with 3D imaging data
    Akula, Manu
    Lipman, Robert R.
    Franaszek, Marek
    Saidi, Kamel S.
    Cheok, Geraldine S.
    Kamat, Vineet R.
    AUTOMATION IN CONSTRUCTION, 2013, 36 : 1 - 15
  • [45] A 2D/3D multimodal data simulation approach with applications on urban semantic segmentation, building extraction and change detection
    Reyes, Mario Fuentes
    Xie, Yuxing
    Yuan, Xiangtian
    d'Angelo, Pablo
    Kurz, Franz
    Cerra, Daniele
    Tian, Jiaojiao
    ISPRS JOURNAL OF PHOTOGRAMMETRY AND REMOTE SENSING, 2023, 205 : 74 - 97
  • [46] A Novel Squeeze-and-Excitation W-Net for 2D and 3D Building Change Detection with Multi-Source and Multi-Feature Remote Sensing Data
    Zhang, Haiming
    Wang, Mingchang
    Wang, Fengyan
    Yang, Guodong
    Zhang, Ying
    Jia, Junqian
    Wang, Siqi
    REMOTE SENSING, 2021, 13 (03)