Suggestions for organizing an ingest structure

[Anthony Weaver]

We are working on a search index related to images captured from trail cameras.  We have an initial structure in place and working that captures metadata like location, and what was detected (animal, person, animal and person).  

We just recently added some additional processing that now includes species classification such as deer, turkey, etc.) and we'd like to add this new information to our metadata structure something like classification: deer.

Where we're stuck is that sometimes there are multiple and different classifications in the same image and I'm not sure the best way to add both into the search index.  For some reference here our subject is the image file and the content includes, detection, image_date, and location and so if an image (or subject) has multiple classifications I'm not sure how to add both entries since they point to the same subject.

Thank you for your help

[Max Tuecke]

Hi Anthony,

Thanks for reaching out. You have two options to solve your issue:
Option 1: Search does support creating multiple entries per subject by setting the "id" field during ingest (see the ingest docs for details). With this approach, determining if an image has multiple classifications could be done by using the subject show API and checking the number of entries. Note that Search only allows up to 10 entries per subject and you would also see all of the entries in your search results.

Option 2: Use an array for the classification field (and any other multi-valued fields). Unfortunately with this method, there is no way to directly query if an image had multiple classifications, so I’d recommend also including a classification count field in the entry.

Let me know if you have any questions.

Best,

Max

[Anthony Weaver]

Max,

Thank you for the response.  I want to be able to include the multiple entries in search(we setup one of the Github based data portals with search enabled)

 so I will give option 1 a try.

Tony Weaver