Building Footprint Generation Using Improved Generative Adversarial Networks
This work addresses the challenge of creating building footprints for urban 3-D reconstruction, which is important for urban planning and mapping applications, but it appears incremental as it builds upon existing GAN methods.
The authors tackled the problem of automatically generating building footprints from satellite images by proposing improved generative adversarial networks (GANs) with a Wasserstein distance-based cost function and gradient penalty. Their method significantly improved generation quality compared to existing networks like conditional GANs and U-Net, while nearly eliminating hyperparameter tuning.
Building footprint information is an essential ingredient for 3-D reconstruction of urban models. The automatic generation of building footprints from satellite images presents a considerable challenge due to the complexity of building shapes. In this work, we have proposed improved generative adversarial networks (GANs) for the automatic generation of building footprints from satellite images. We used a conditional GAN with a cost function derived from the Wasserstein distance and added a gradient penalty term. The achieved results indicated that the proposed method can significantly improve the quality of building footprint generation compared to conditional generative adversarial networks, the U-Net, and other networks. In addition, our method nearly removes all hyperparameters tuning.