Visual matcher - not working? not sure

Support
,
1

What Wildbook are you working in? ACW

What is the entire URL out of the browser, exactly where the error occurred?
It may not be an error but I’m just not sure I understand how the Visual Matcher is supposed to work. I’ve tried to use visual matcher a few times and the results have not been helpful; in fact I believe that on a previous attempt (did not keep the details unfortunately), the Search filter options did not produce expected results and it appeared to be a bug at that time.

Can you describe what the issue is you’re experiencing?
How does “Visual Matcher” work? How does it select possible matches and how is it different from the automated matching process?

Can you provide steps on how to reproduce what you’re experiencing?
I used the encounter record below and selected “Visual Matcher” from the image with the bounding box around the dog’s body. I also selected “pigmentation” as “tan”.
Wildbook for Carnivores | Login

I received the following results which included at least one "dark’ dog. It also included viewpoints that don’t match the viewpoint of the dog in the target image but I assume that’s because those keywords are not used in this process?

If this is a bulk import report, send the spreadsheet to services@wildme.org with the email subject line matching your bug report

2

Hi @ACWadmin1,

Visual Matcher is not likely a as you good choice for social species. Under the covers, it is a limited Encounter Search that displays each Encounter as a result row filed with its photos (not annotations). It also does not use the photo keywords for dog color but rather values from the legacy Encounter.patterningCode database field.

In less social species, where there is generally one animal per image, this tool still holds some value (e.g., MantaMatcher.org). But for social species and those benefiting from ML-based detection, an entirely new visual matcher is being built to add more search filters (e.g., photo keywords) and focus the eye on annotations.

Thanks,
Jason

3

Thanks Jason. It’s good to understand that we shouldn’t bother with Visual Matcher for our social species until the new version comes around. I could see that new version being popular with some researchers who would like to be able to do that themselves.

cheers
M