Thanks for your comment, @wasserth. I believe the weights are just at the “TOTALSEG_WEIGHTS_PATH” independent of 3D Slicer.
In Windows 11 I find the “3d_fullres” folder only once at
C:\Users\Rudolf\.totalsegmentator\nnunet\results\nnUNet\3d_fullres\
after a successful TotalSegmentator install.
thank you very much
and I have just find the path like you described and I run the source TotalSegment code in linux and the path was ’ :~/.totalsegmentator/nnunet/results/nnUNet/3d_fullres’.
and I have just put the weight zip files unzip there. the code run well , and the results were amazing
thank you.
5 posts were split to a new topic: TotalSegmentator error at first run: ‘DummyFile’ object has no attribute ‘flush’
That’s right! Thank you very much!
Hello,
I want to now if totalsegmenter goes fast with mac book M2, and which modules does it requere?
Thanks
AI support in Apple Silicon systems is still lagging behind. Apple MPS acceleration is available now for some pytorch features, which significantly improves speed. However, MPS-accelerated pytorch is not usable for TotalSegmentator yet, because some features required by nn-UNet are not supported.
Generally Slicer on mac systems works pretty well most operations, however ML models aren’t as well supported since most of that work requires Nvidia systems for top performance currently. That may change in the future, but that’s the state of things now.
