Abstract: Our study proposes a new approach for building extraction by introducing the Deformable-Convolution Fusion Feature module, based on the Dual-Attention Network and deformable convolution. The ...
The encoder employs a DenseNet-B (bottleneck) architecture with three dense blocks separated by transition layers. Each bottleneck layer consists of a 1x1 convolution (expanding to 4x growth rate) ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results