About
Solution Catalog
Documentation
Tutorial
cellcanvas album catalog
sharing cellcanvas tools
morphospaces
/
train_swin_unetr_pixel_embedding_mrc
/ 0.0.1
Train SwinUnetr Pixel Embedding Network with Mrc Dtataloader
Train the SwinUnetr pixel embedding network using the provided script and dataset.
Tags
imaging
cryoet
Python
napari
Citation
Morphospaces team.
https://github.com/morphometrics/morphospaces
Solution written by
Kevin Yamauchi
Kyle Harrington
Zhuowen Zhao
License of solution
MIT
Source Code
View on GitHub
Arguments
--pretrained_weights_path
Pretrained weights path (default value: PARAMETER_VALUE)
--train_tomogram_path
tomogram mrc files directory for training dataset (default value: PARAMETER_VALUE)
--train_mask_path
mask mrc files directory for taining dataset (default value: PARAMETER_VALUE)
--val_tomogram_path
tomogram mrc files directory for validation dataset (default value: PARAMETER_VALUE)
--val_mask_path
mask mrc files directory for validation dataset (default value: PARAMETER_VALUE)
--lr
Learning reate for the supervised contrastive learning (default value: 0.0001)
--logdir
output dir name in curr working dir. Default is checkpoints (default value: checkpoints)
Usage instructions
Please follow
this link
for details on how to install and run this solution.