What process are you using to start a match? I can see the result you’re seeing, but every time I run “start another match”, I get the expected annotation.
Has this annotation:
which is indeed wrong.
However, I am seeing that annotation as the one displayed in the result after I click “start another match”:
Are your matches being kicked off by the bulk import or a different button?
Hi @jason - we’re only using the start another match function; we’re not deleting and re-bulk uploading these encounters so we can’t kick them off from the bulk import side. Also, if it was being done that latter way, we would see all annotations from that single media asset displayed on the same match results page, which we’re not seeing. Regardless, these matches were all kicked off using start another match.
The problem is, the result of incorrect encounter as the target image doesn’t happen every time. Sometimes within the same sighting, I’ve seen some multi-annotation media assets produce the correct annotation as the target on the results page and some multi-annotation media assets produce the incorrect annotation as the target on the results page. So the fact is, you’d have to do as Paul’s doing, re-run matching for multiple encounters to start seeing this issue. But you can see from each match results page that all these sightings for EWT’s wild dog census were re-run again yesterday after you told Paul about the fix you’d deployed.
The sightings he re-ran matching on yesterday are all those found this way:
Search > Sighting search > under “Collected Data Fields”, select Lycaon pictus twice (don’t use the one with the underscore), and enter “census” in the Occurrence ID field, then under Metadata filters, under “Assigned to user” select Grant Beverley.
Paul has re-run matching on all encounters, via “start another match” in each of those 31 sightings several times this past week and continues to find multiple instances of this issue.
Hope that helps,
Interesting but unlikely to be related issue: I just noticed that the “assigned to user” list under Sighting search is a list of user’s full names but the “assigned to user” list under Encounter search is a list of user IDs. Both appear to give the correct results regardless but I thought I’d point it out. Never know what might be related…
Asking @jon and @MarkF to help here to provide a second and third set of eyes on the issue.
Thanks @jon & @MarkF
This sighting is a good example of the issues we encounter.
Our workflow is to search for a sighting, and then right click on each encounter to open in new tabs.
Then we use the Zebra menu on each body encounter to check match results, or run matching again.
Results show as follows.
One body annotation in image - match results show that animal.
Two or more annotations in image - match results consistently display wrong annotation on most, but not all encounters.
Tall portrait images - match results do not even show another animal - typically some patch of ground or grass. (See Bug report here: Annotation offset in match results - matching against grass rather than animal )
Example in this sighting:
Thanks @ACWadmin1 @PaulK for confirming the fix for ticket WB-1691 worked!
Thank you all for sticking with it until you found the problem and fixed it.
@jason @jon @MarkF
Thanks to all for figuring this one out! I tested a cheetah example that I had and it looked fine, as well as some wild dog examples. All appear good!