In which Wildbook did the issue occur?
Internet of Turtles
What operating system were you using? (eg. MacOS 10.15.3)
Various, e.g. MacOS 15.1.1
What web browser were you using? (eg. Chrome 79)
Various, e.g. Chrome 133.0.6943.55
What is your role on the site? (admin, researcher, etc)
OrgAdmin, Researcher (Olive Ridley Project)
What happened?
Automatic annotations are working again (yay!) and matching is also initiated, but the algorithm seems to match against the wider region at least, not taking into account the Location ID of the encounter.
This leads to matching against 28500+ candidates instead of 2800. It takes the system 48hrs+ to return a match for a single encounter like this - maybe the matching against such a large number of candidates makes it so slow?
Also, lack of regional restrictions is not really helpful at the moment - we’ve barely seen turtles move between atolls in the Maldives, let alone entire countries/oceans.
What did you expect to happen?
Automatic matching to be specific to the Location ID provided.
What are some steps we could take to reproduce the issue?
Upload an encounter and check what individuals/encounters the automatic matching process is using.
Please let me know if you need further examples. I have only included this one (ultimately unidentifiable) encounter in here because it illustrates the principle. For other recent uploads, I have manually rerun the matching process for at least one side of the face - that was always automatically set to the correct location.
Quick update: this code hasn’t changed in a long time, so it’s likely that this behavior has been going on unnoticed for a while. I’ll need to write up a ticket for it and I’ll post a link to it here when it’s ready.
Just a quik follow up to check if there has been some development here?
Until yesterday, our workaround (i.e., manually starting a new match for each newly uploaded encounter) did work quite well until yesterday, but today a new issue has occurred as even locations with the correct Location ID assigned return the No data to match against error. This applies for both automatic and manual matching processes.
I don’t have any updates yet, but I do see the fix is being worked on as part of Wildbook 10.7.0, which should be released this month.
I checked the match page examples you shared and it looks like most of the errors are on the head-only annotations. Did the error message only appear when you sent it for re-identification or was the error there during the original match results, too?
Quick update: the missing results are a separate bug related to part matching. We’re working to get this patched quickly. I’ll let you know when that’s ready.