Swin-Unet on 2D images

Question

Swin-Unet on 2D images

Opened this issue 2 months ago · 2 comments

Is it possible to use Swin-Unet on 2D images for different types of segmentation tasks? And if so, how should I proceed?

Answer 1 · 2024-06-27T14:16:07.000Z

Is it possible to use Swin-Unet on 2D images for different types of segmentation tasks? And if so, how should I proceed?

Do you have a solution? I also need the 2D version of Swin-UNet for experiments.

Answer 2 · 2024-07-02T10:35:19.000Z

Hey, I managed to run the code but all my predictions were plain black. I tried to allocate one single image to the .nii.gz file, but something is still wrong.

I changed the model for my segmentation task, which are better for 2D images. Maybe try one of them:
nnUNet, UNet++, Finetune-SAM, SAAB