Stats gen fix
#8
by
egrace479 - opened
Noticed (while making stats-gen script) that the stats generation process here used a fixed length of catalog instead of calculating it. It was fixed at the original number (not updated with v3.3), so updated the cell to use the number of images (length of catalog without duplicated train_small), and updated the associated stats CSV.
@thompsonmj , look good?
egrace479 changed pull request status to
open
Properly updated now. It doesn't generate the standard deviation for genus or species. The common average and standard deviation are a better measure of this because species names are duplicated across various genera (there are only about 163K unique instances of species, despite 454K unique 7-tuple taxa and 443K unique common instances).
Great!
egrace479 changed pull request status to
merged