UPD: Update HoVer-Net+ Post-processing

[adamshephard]

Update an issue with the HoVer-Net+ post-processing. Previously, a simple blur was applied for the post-processing. Now it is updated inline with the original work - using various morphological operations specific to each epithelial layer.

Remove warnings from importing specific libraries in both HoVer-Net and HoVer-Net+.

[codecov[bot]]

Codecov Report

Merging #431 (17feb77) into develop (bc5ee15) will increase coverage by 0.00%. The diff coverage is 100.00%.

@@           Coverage Diff            @@
##           develop     #431   +/-   ##
========================================
  Coverage    99.65%   99.65%           
========================================
  Files           61       61           
  Lines         6064     6082   +18     
  Branches       995      997    +2     
========================================
+ Hits          6043     6061   +18     
  Misses           9        9           
  Partials        12       12           

Impacted Files	Coverage Δ
tiatoolbox/models/architecture/hovernet.py	100.00% <100.00%> (ø)
tiatoolbox/models/architecture/hovernetplus.py	100.00% <100.00%> (ø)

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

[John-P]

Looks good. Just a small note on performance, though. I think the cv2 equivalent operations might be a bit faster than the skimage ones. This is minor but could have a bigger impact on a WSI level.

See https://opencv24-python-tutorials.readthedocs.io/en/latest/py_tutorials/py_imgproc/py_morphological_ops/py_morphological_ops.html#morphological-ops

[shaneahmed]

Looks good to me. I agree with John's comment on using opencv for morphological processing.

[adamshephard]

Looks good. Just a small note on performance, though. I think the cv2 equivalent operations might be a bit faster than the skimage ones. This is minor but could have a bigger impact on a WSI level.

See https://opencv24-python-tutorials.readthedocs.io/en/latest/py_tutorials/py_imgproc/py_morphological_ops/py_morphological_ops.html#morphological-ops

Hi John, after looking at this I see that there are no opencv equivalents for the functions that I have used. I've also looked at the source code for these functions, and they rely on scipy anyway. Probably best to keep as it is?