In which Wildbook did the issue occur? ACW
What operating system were you using? Win 10
What web browser were you using? latest chrome
What is your role on the site? admin & researcher
We ran matching for this encounter: https://africancarnivore.wildbook.org/encounters/encounter.jsp?number=641ff3e8-73d6-4ba0-8588-e283c8e978b2
We got these match results: Wildbook for Carnivores
Proposed match #1 in the list is a duplicate of the target image - same date, time and we can see from the sighting info of the target image that this was the only left profile image in that sighting. Clearly it was deemed a good ID pic and so the original researchers curating these images selected that photo and used it as the left profile ID pic for CH20’s ID kit, thus the duplication.
What confuses me is why the match image doesn’t have the tag of “alternate ID” on it? Is it because the “alternate ID” pic is actually the target encounter?
What did you expect to happen? I expected to see a bubble on either the target or the proposed match that shows the “alternate id” duplication indicator, if that makes sense.
Apologies, I’m not sure I understand the question. What value for alternateID did you expect to see in this example?
Sorry for not being clear - usually there’s a text box overlaid on an image in match results that has an exact duplicate in the system, with a link to that duplicate encounter included. Maybe the terminology in that box is alternate encounter not alternate id??? Sorry, I can’t remember what it usually says.
I found an example from a previous support request here: What does the "alternate references" msg on target in match mean?
That makes sense.
While they look identical and are likely from the same images, the underlying annotations and images are in fact not identical. By clicking on the query and target Encounter, I can see that one MediaAsset (target)
looks like it is a cropped version of the query image:
This means that Image Analysis sees them as two separate images (they get different hashed acmIDs) and detection works slightly differently on each, causing non-duplicate annotations to be created (also different acmIDs on the annotations).
Without true duplication (matching acmIDs on the Annotations themselves), we cannot propose alternate IDs because we cannot verify that they are truly duplicates.
I see that now; should have figured it out myself. Sorry for the trouble, Jason and thank you.