Model monomerization method and system based on deep learning

The invention relates to a model monomer method and system based on deep learning, and the method comprises the following steps: obtaining a scene image data set, and generating a three-dimensional model of a scene through a three-dimensional reconstruction method; obtaining a required vector map shp file; elevation information of the building is obtained from the shp file of the vector map, and according to the height information, the planar building model is pulled up into a three-dimensional model through extrusion operation. According to the method, the monomerized model is generated through a deep learning neural network CEDN + OfficientPS algorithm, and a method of manually extracting a model contour does not need to be adopted. 本发明涉及一种基于深度学习的模型单体化方法及系统，其包括如下步骤：获取场景图像数据集，且通过三维重建方法生成场景的三维模型；获取所需要的矢量地图shp文件；从矢量地图shp文件中获得建筑物的高程信息，且根据所述高度信息，将平面的建筑物模型通过挤出操作拉升为立体模型。本发明通过深度学习神经网络CEDN+EfficientPS算法生成单体化模型，无需采用人工提取模型轮廓的方法。