search

uni-medical / SAM-Med3D

SAM-Med3D: An Efficient General-purpose Promptable Segmentation Model for 3D Volumetric Medical Image
Apache License 2.0
427 stars 56 forks source link

possibility for multiclass semantic segmentation #34

Closed ramdhan1989 closed 4 months ago

ramdhan1989 commented 6 months ago

Hi, I would like to ask if the code is easily to be modified for multiclass semantic segmentation?

please advise,

thanks

blueyo0 commented 5 months ago

I think it is easy. You only need to change the seg head of the model to support multi-class seg and modify the training pipeline.

However, I don't suggest you to do som because this code is only experimental code for research of promptable seg. A better way is to modify a SOTA multi-class semantic seg code (e.g. STU-Net，nnU-Net), and then you only need to change the model.

Hope this can help you.