What Wildbook are you working in? Wildbook for Leopard Sharks
What is the entire URL out of the browser, exactly where the error occurred? https://leopardshark.wildbook.org
Can you describe what the issue is you’re experiencing? I have processed over 100 leopard shark IDs to from Cape Byron to date but 95% of the time, when I process new IDs it doesn’t bring up a match (see attached). I’m not sure the matching is working properly?
Can you provide steps on how to reproduce what you’re experiencing? Run a new ID and no matches will be made.
If this is a bulk import report, send the spreadsheet to email@example.com with the email subject line matching your bug report
A scan task that had upwards of 80 spots has caused the platform to hang. We are looking at how to prevent this from happening in the future. I have restarted the system for you, but existing jobs will need to be manually restarted from the scanTask admin page. Any task that has greater than 80 spots needs to be manually annotated.
Was this me that had the 80 spot ID? Was it in my list of pending IDs? I don’t believe I would have used 80 spots in one ID? If you could point this out to me that would be great.
It is still stuck, nothing is going through, even new ones that I put through.
I think the issue is still there.
We’ve been doing some investigation on this. It seems that the slow processing is indeed due to long comparisons between images with many spots. In most spot mapping platforms we see an average of around 30-40 spots per image, and that is enough to have accurate matching results and a pretty quick turn around.
Because of a larger matching area and plenty of spots on the species we are seeing more on average- the maximum I saw for an image on this platform was 118.
I didn’t see 80+ spots on any encounters managed by your project, but here is one with 77:
I notice a couple in the 60’s as well. We’ve added some extra computer power to get through the existing queue and I am seeing progress, though it is still moving somewhat slowly.
To ensure shorter turnaround times, we have added a limit of 50 on the amount of spots that can be mapped for a given image. Based on some optimization work we have done in the past this seems to be around where compute times start to quickly extend and accuracy doesn’t improve much with the addition of more spots.
This won’t effect existing mapped images. I’d encourage re-mapping images with spot counts over 50 if you find them.
Hope this helps, we’ll continue monitoring the progress of the existing match queue.
Thanks for your response.
The system seems to be stuck again with 1 comparison to go on a couple of IDs; see attached image; it has been like that all day.
If you can help push it through that would be great.
Sorry for the delayed response. It looks like jobs are moving through the queue now,
are you seeing results?
I’ve monitored one job of yours with 38 spots through to completion and see it starting to work
on another with 43.
They are moving through much faster.
I also found that if you wait for the list to complete and then process more IDs it works through them faster. If you add more IDs whilst it’s still processing previous ones, this seems to bump the IDs being processed to the bottom of the list and it sticks.