3 ms·
Could it be that this old, abandoned image processing software glued with Python is what you're expected to improve/modernize? First you have to play with it t
by kazinator 5y ago
Could it be that this old, abandoned image processing software glued with Python is what you're expected to improve/modernize?
First you have to play with it the way it's being done now.
Research places always have legacy cruft; it's not like they just heard of image processing and want this new thing now.
How I would approach it would be to focus on reproducibility of old results. OK, you have the legacy cruft, you have Python and you have data. Put that together to create a regression test suite which validates that the processing produces certain results.
Once you have that, you can confidently replace legacy parts of it with new stuff, and show, "look, the tests pass; it's handling the existing data in the same way".