This seminar presents an overview of building extraction from images, highlighting its importance for urban planning, disaster management, and 3D city modeling. Various data sources such as optical imagery, SAR, and LiDAR are discussed, along with methods ranging from pixel-based and object-based techniques to machine learning and deep learning approaches like CNNs and U-Net. The seminar concludes that hybrid methods combining multiple data sources and algorithms provide the most accurate results, which are essential for reliable building modeling and LOD2 3D reconstruction.
Dr.haval abduljabar
