search

TFboys-lzz / MPSCL

This repository contains code for the paper "Margin Preserving Self-paced Contrastive Learning Towards Domain Adaptation for Medical Image Segmentation", published at IEEE JBHI 2022
29 stars 1 forks source link

train on another dataset #5

Closed wh-isper closed 1 year ago

wh-isper commented 1 year ago

Hi author! I have a doubt about using your code to train on another dataset. At first, I need to pre-train a segmentation model on the source domain to obtain the initial category prototypes. Is it necessary to pre-train main and aux discriminators?

wh-isper commented 1 year ago

Besides, in ct2mr.yml and mr2ct.yml, you choose different models for D_type(original_d and patchGAN), which one should i choose?

TFboys-lzz commented 1 year ago

Hi wh-isper: If you just want to get the initial category prototypes, the discriminators are not required to be pre-trained. About the selection of the discriminator, you can verify it on your own dataset. We found experimentally here that these two discriminators have their own advantages in two different directions

wh-isper commented 1 year ago

Thanks for your quick reply!