Extraction buildings from very high-resolution images with asymmetric siamese multitask networks and adversarial edge learning

Xuanguang Liu, Yujie Li, Chenguang Dai, Zhenchao Zhang, Lei Ding, Mengmeng Li, Hanyun Wang

Abstract: Highlights•We proposed an asymmetric Siamese multitask network with a dense, high-level feature fusion strategy for synergizing CNNs and Transformers.•We constructed a building-edge constraint technology to improve the geometric accuracy of the results.•We designed a pseudo-sample generation strategy based on a self-training paradigm and building-instance transfer.