Closed nuwanarti closed 4 years ago
Based on your outputs I am expecting that you are running the example notebook.
You said that you are stopping the process. Looks to me from your first print out that you are just missing three lines from the output of what I expect:
time to linear sum assignment: 7.1949567794799805
Number of unique correspondences: 5000
Number of unique correspondences after smoothing: 4713
The values in there don't matter as much as the fact that those lines should be printed. This to me means that you arent finishing the linear_sum_assignment which is the hungarian algorithm described in the original paper (line 275 in pyfocusr.py). This is what is used to find correspondences between the meshes at the end of the registration. This makes sense why you are then able to run the next cell that plots the aligned spectral coordinates (because this part has completed - align the spectral coords), but then it fails when you try to run the other cells that include visualizations of the aligned meshes, which require correspondences to have been found and then new meshes to be generated from the correspondences.
Based on your print outs of how long things took, your registration is running faster than on my computer, and from my print out the linear_sum_assignment is only 7ish seconds. So, I would expect that it should occur pretty quickly. Did you let it run for some time? Without letting that part run/complete you won't get aligned meshes and there will be nothing to display in the second cell that is crashing, so those errors don't surprise me.
So... I guess my first suggestion is to just let it run for a while. If it is taking considerable time, I wonder whether there is some other bottleneck, your computations are faster than mine, maybe it is ram? How much ram do you have? Im running that step on a 2015 laptop with 16gb ram, so nothing crazy.
Lastly, have you changed any of the other default parameters in the example?
I let it be for 2 hours but still it didn't complete the process
Please see my comment on the other issue (https://github.com/gattia/pyfocusr/issues/2). I updated some requirements and the build instructions on the main readme. This should mean that if you run the example notebook that everything should be identical to my setup. Hopefully this resolves any issues. If it still persists, if you could start by confirming what code you are running and how you are running it, what environment, your dependencies, and a script to reproduce the error that would help.
Thankyou very much its working now on mac os catalina
Excellent.
Hope that it ends up serving your needs.
Looks like reg.align_maps() won't complete the process. here is the output, but it doesn't complete the process
if I stop the process and continue then I'm getting the following error
TypeError: Could not find a suitable VTK type for object
while executing below linereg.view_meshes_colored_by_spectral_correspondences()
complete error is as follows