low training speed on S3DIS?

MIC-DKFZ / MedNeXt

[MICCAI 2023] MedNeXt is a fully ConvNeXt architecture for 3D medical image segmentation.

https://arxiv.org/pdf/2303.09975

Apache License 2.0

345 stars 26 forks source link

low training speed on S3DIS? #17

Closed tangze5258 closed 4 months ago

tangze5258 commented 8 months ago

every epoch took about 6000 s , somgthing wrong?

tangze5258 commented 8 months ago

how to predict single *.nii file if the model trianing ends?

saikat-roy commented 6 months ago

Hey @tangze5258. Sorry for the late reply. You can use the following mednextv1_predict instead of nnUNet_predict as in the example here for inference on single files.

I have never trained on S3DIS but the training speeds depend heavily on the CPU, type of storage the training data is on (HDD, SSD etc) and the GPU used. If you could tell me about your configuration, I could confirm whether this indeed is the issue.