有办法只输入一张图像，然后提取建筑吗，不做cd检测

[kuaiqushangzixiba]

如题

[ABCnutter]

You might be referring to semantic segmentation tasks. You can refer to the mmsegmentation repository (https://github.com/open-mmlab/mmsegmentation) for more information.

[kuaiqushangzixiba]

I know what you mean, I have a need, sometimes semantic segmentation, sometimes change detection, so I want to use one algorithm to achieve 2 functions