Semantic segmentation on big images

[KFrank]

Hi Sylvain!

@Sylvain_Ard

Hello noone has an answer please ?

I suspect that people tend not to answer your questions because you ask
the same question over and over again without seriously engaging with the
answers you were given when you asked it earlier.

To summarize and answer I gave you previously, U-Net is designed to work
on images of essentially any size using the tiling strategy discussed in the
original U-Net paper.

Here is my earlier answer:

Semantic segmentation on very big images

Hi Sylvain! One reasonable approach would be to use a U-Net (a common and solid semantic-segmentation network architecture) with the so-called “overlap-tile strategy” (that is, apply the U-Net to patches of the large input images). This is discussed in detail in the original U-Net paper. Best. K. Frank

Best.

K. Frank