What is the entire URL out of the browser, exactly where the error occurred? Flukebook Flukebook
Can you describe what the issue is you’re experiencing?
Within my project I cannot seem to access my match results. It was working fine yesterday but I haven’t been able to display any matches today. The page has been displaying ‘Waiting for results.’ for multiple hours today (despite reloading the page, waiting, logging out and back in, etc.) and occurs on any encounter match I try to look at within my project. Any advice would be helpful, thank you
Can you provide steps on how to reproduce what you’re experiencing?
See links. Trying to open any match results within a project will result in the same ‘Waiting for results’ message and no results showing.
We had >2000 machine learning jobs go through the pipeline this weekend. It looks like your ID jobs were lost in that queue. I just pushed them through again (422 ID jobs between the algorithms), and I’ll post back here once they’re done. They’re at the top of the queue, so they are already being processed.
Thank you! Yes that does seem to have resolved my problem, I can now see the results of the matching for that project.
An additional question: I have another project that is showing ‘attempting to fetch results’ for one of the algorithms and ‘0 - inspection image unavailable (likely outdated)’ [see image] for my matches on the other algorithm despite only sending these for ID a few days ago, before my previous issue started.
Yes, please. Let us know if you see anything unusual with the updated match results.
If you’re referring to viewing new matches, you can do this by re-running your match results. If you mean that you’ve already approved previous match results and want to start over again, you can edit the Identity section of the Encounter page and click the button next to Remove from marked individual. You can then start a new match.
I have resent the encounters to be matched and it looks like the algorithms have been successful, however I can not physically see the matches, only a blank space (see image below).
I’ve refreshed the page, logged back in and out and checked multiple encounters and this same error is on all the match results. Is there anything that I can do or will these need to be resent to ID again?
Can you please send the URL from your browser to the screenshot? I clicked around, and anything that has been reesent recently is displaying match results for me. I’d like to look at the example you see.
Here’s one of the links to the encounters, apologies, I can’t remember exactly which encounter I used for the screenshot above but the encounters are all still presenting this problem for me.
I’m glad I’m not going crazy and that you can see the problem too!
When I click on that second link I can see the results of the matching, although I am still getting the 'inspection image unavailable (likely outdated) message for the CurvRank algorithm.
I had 2 large bulk imports for this project and so I just resent the entirety of the import back through ID from the respective bulk import pages (see links below)
Those CurvRank scores are all zeroes, meaning no evidence of match. We don’t generate the trace image for CurvRank when the score is zero. You can interpret all 0 scores from CurvRank as " no match was found".
I resent your first bulk import to ID, and when it finished, I will send the second. The new results look good and should not have that empty content.
Thank you for clarifying this as I was not aware this was how the output for the algorithm worked. I guess I am only concerned because this algorithm worked well for my other catalogue (Left dorsal side) and therefore I am confused why it wouldn’t be working for the Right dorsal side on any of the encounters.
Great question. I can say that the algoithm considered 819 other fin annotations (bounding boxes). Nothing matched.
It looks like the individual has no other right- or left-side fins assigned to it currently, so the algorithm had no opportunity to match against ground truth data and found 0 false positives, which would be scores greater than 0. That’s generally a good thing as long as there isn’t a match to this individual hidden somewhere else in the dataset.
