Pipeline for particle picking in cryo-electron microscopy images using convolutional neural networks trained from positive and unlabeled examples. Also featuring micrograph and tomogram denoising with DNNs.
The training metrics written to the results.txt file are very useful to look at graphically (see #78). I already have some code to plot them (https://github.com/Guillawme/topaztrainmetrics/).
Would you be interested in integrating this plotting code into topaz, to make it write png files with these plots by defaults, along with the results.txt file? The only additional dependency that topaz doesn't already require is matplotlib. It would make topaz more user-friendly, and give users a more direct feedback on their training results.
If you are interested, I am willing to prepare and submit a PR, but I would need some guidance: in which file should I add this code (I guess wherever the dataframe containing what is eventually written to results.txt is generated, but I couldn't find where that is).
Hello,
The training metrics written to the
results.txt
file are very useful to look at graphically (see #78). I already have some code to plot them (https://github.com/Guillawme/topaztrainmetrics/).Would you be interested in integrating this plotting code into topaz, to make it write png files with these plots by defaults, along with the
results.txt
file? The only additional dependency that topaz doesn't already require is matplotlib. It would make topaz more user-friendly, and give users a more direct feedback on their training results.If you are interested, I am willing to prepare and submit a PR, but I would need some guidance: in which file should I add this code (I guess wherever the dataframe containing what is eventually written to
results.txt
is generated, but I couldn't find where that is).