Enable capture of image recognition metadata

[kitenetter]

Jim tells me that image recognition metadata is being captured within the apps for EBMS and e-surveyor, but not within the iRecord app. We would like to capture this data from the iRecord app as well.

I'm hoping that it will be relatively straightforward to apply the process from the other apps that are already doing this. Can you let me know how big a job this is?

[kazlauskis]

@kitenetter It's a small task, I can do this as part of #237.

[kazlauskis]

@JimBacon could you please update the IndiciaAI proxy module on the iRecord site? I need the classifier to process multiple images per request, similar to what we did for the e-Surveyor.

[kazlauskis]

@kitenetter One unclear area is how to treat a situation where 1) users add several photos 2) use the classifier for suggestions and 3) remove or change those photos. It would be technically incorrect to send the user preference metadata with the new set of photos to the database because the suggestions were different when the user has decided to select the species. Is it OK to not send any classifier metadata where photos were modified or reclassified after the original user decision?

[kazlauskis]

This requires this warehouse ticket: Indicia-Team/warehouse#534

[kitenetter]

@kazlauskis asked: "Is it OK to not send any classifier metadata where photos were modified or reclassified after the original user decision?"

I think we'll have to do that (i.e. not send metadata in such cases) - I can't think of any other practical way of dealing with the complexities of multiple photos and multiple subsequent updates to the ID decisions.

[johnvanbreda]

I've fixed Indicia-Team/warehouse#534.