Begin building annotation, training & testing dataset + artifacts

[Jesssullivan]

Beginning to fill in preprocessing steps to construct a sample training & testing dataset + json artifacts for a Mushroom Observer annotator tool.

[Jesssullivan]

Looking forward to feedback, thoughts, ideas & directions! This will hopefully help us get started.

• Notes on this PR here

Recently discovered @Olament's fun mycology projects as well. Any interest in integrating with & expanding this neat project MycologyBot with MO data? I am hoping to poke around and fiddle with the DeepMushroom & MycologyBot repos this weekend.

todos:

• Add MO annotation tool!
• Experiment with training lineages described by Danish Mycology Society- V1, V2)

[Jesssullivan]

ok, now the web annotator is working over here and can process Mushroom Observer annotation tasks generated by python3 preprocess.
To try this out, snag an annotation task:

# ...along the lines of:
wget https://mo.columbari.us/static/mo_example_task.tar.gz
tar -xf mo_example_task.tar.gz

...then select this directory to upload at the annotation landing screen.

(The vague idea here was to maybe recycle some existing Merlin AI model training stuff, I think @gvanhorn38 was maybe hoping to add more variety and examples for future projects just like this one)

More notes here:
https://jesssullivan.github.io/mo-image-identifier/

[Jesssullivan]

alright, added the methods for @mo-nathan 's second csv format. the above annotation json / artifacts / images @ mo.columbari.us have been updated too with the 900 image dataset.
Many more notes and whatnot added here too:
https://jesssullivan.github.io/mo-image-identifier/