Hi @zhanghaicang ,thanks again for sharing this code.
I'm also trying to reimplement his code now.
However, I'm having a memory issue here, 12GB of GPU memory is never enough for me.
Operating directly on a L*L map with no pooling looks unacceptable.
Any advice?
Hi @zhanghaicang ,thanks again for sharing this code.
I'm also trying to reimplement his code now.
However, I'm having a memory issue here, 12GB of GPU memory is never enough for me.
Operating directly on a L*L map with no pooling looks unacceptable.
Any advice?