Hi Paul. I thought this place makes it a bit easier to explain my idea.

PHASE 1: to be done by myself

I will take all your pictures, and for each of them I will extract a few ¼” areas and some ½” areas:

I will end up with a very long package of ¼” pictures and ½” pictures (something like your hobbithouseinc.com/personal/woodpics/_anatomy/anatomydump.htm, except for the fact that in one category they will ALL be ¼” and in the other one they will ALL be ½”).

I expect to end up with roughly 5×4500 ¼” pictures and 3×4500 ½” pictures.

PHASE 2: to be done by Plant-net team

They will put all the pictures in their algorithm. I don’t know how it works, but MAYBE it works as good for endgrains as it does for leaves. And it does work strikingly well.

PHASE 3: to be done by users (including me!!!!)

The user will upload an endgrain picture.
Then he will be asked: “is it a ¼” section, or is it a ½” picture”;
based on this information the software will look for a match in the ¼” “folder” or in the ½” “folder”.

Then the user will receive something like this:

Juglans – 35%
Quercus – 4%
Catalpa – 1%

It will absolutely not be an ultimate and conclusive answer, but it may point the user in the right direction for further checks.

I can tell you for sure that I would use it VERY often!

I understand if you are not up for this, but I believe scales are not “bastardized” by this process; it is rather a way to narrow down the search in a quick way.