NEW: Add NuClick architecture

[mostafajahanifar]

This PR adds NuClick architecture, as described in the paper below, to the toolbox.

Koohbanani, Navid Alemi, et al. "NuClick: a deep learning framework for interactive segmentation of microscopic images." Medical Image Analysis 65 (2020): 101771.

[codecov[bot]]

Codecov Report

Merging #449 (1c678e9) into develop (a90b330) will increase coverage by 0.00%. The diff coverage is 100.00%.

@@            Coverage Diff            @@
##           develop     #449    +/-   ##
=========================================
  Coverage    99.63%   99.63%            
=========================================
  Files           61       62     +1     
  Lines         6240     6379   +139     
  Branches      1035     1043     +8     
=========================================
+ Hits          6217     6356   +139     
  Misses          10       10            
  Partials        13       13            

Impacted Files	Coverage Δ
tiatoolbox/models/architecture/idars.py	100.00% <100.00%> (ø)
tiatoolbox/models/architecture/mapde.py	100.00% <100.00%> (ø)
tiatoolbox/models/architecture/micronet.py	100.00% <100.00%> (ø)
tiatoolbox/models/architecture/nuclick.py	100.00% <100.00%> (ø)
tiatoolbox/models/architecture/sccnn.py	100.00% <100.00%> (ø)
tiatoolbox/models/architecture/unet.py	100.00% <100.00%> (ø)

:mega: We’re building smart automated test selection to slash your CI/CD build times. Learn more

[mostafajahanifar]

@kesixu @mostafajahanifar

The nuclick architecture file needs to have docstring consistent with the rest of the repo.

It needs to be fixed. The remaining three files are good.

Dear Shan,

Thanks for your comments and commit suggestions. I've incorporated all of them in the PR. Please check if you like the changes.

[mostafajahanifar]

Dear @shaneahmed

I revised the code to make the input to all models consistent with num_output_channels argument. In particular, I changed MicroNet argument to num_output_channels but for MapDe changed it to num_classes because argument is actually referring to number of cell classes based on which number of output channels is generated ( num_output_channels=2*num_classes ).

I have also refactor all the models to use predefined (base class) preproc method instead of _tranform method. And made some other changes to avoid [PYL-W0221] pylint error.

[shaneahmed]

Dear @shaneahmed

I revised the code to make the input to all models consistent with num_output_channels argument. In particular, I changed MicroNet argument to num_output_channels but for MapDe changed it to num_classes because argument is actually referring to number of cell classes based on which number of output channels is generated ( num_output_channels=2*num_classes ).

I have also refactor all the models to use predefined (base class) preproc method instead of _tranform method. And made some other changes to avoid [PYL-W0221] pylint error.

Thanks @mostafajahanifar Please can you check the errors?