Missing annotations - repeat of WB-768?

In which Wildbook did the issue occur? ACW

What operating system were you using? Win 10

What web browser were you using? latest chrome

What is your role on the site? researcher for uploads

What happened? I believe these are additional instances of the bug being tracked under WB-768. I reviewed my other imports from the past couple of weeks and found additional instances of missing annotations from bulk imports that had IA run on them.
Is it possible to try the fix that @jason used the last time - “I echoed the call back result from detection (which was successful), and this time the server accepted it.”
The import IDs for the batches with missing annotations are as follows:

Oddly, there is one image in another bulk upload that I would have expected to see annotations in but there are none. May not be related but I thought I should point it out just in case:


I won’t bother forwarding the import files unless you tell me you need them, since I believe this is a known issue.


Hi Maureen,

I repeated the detection callbacks on these, and I now see bounding boxes.

I’m testing a fix for WB-768 (caused when bounding boxes don’t appear after detection). This is a subtle bug, so this first pass may or may not address it.

Can you please let me know if you see these missing detections again soon? That will let me know if the solution works or not.


Hi Jason, I’ve just had a look at the encounters for one of the sightings that had this problem and have found annotations in the images now (yay!) but a slightly different additional issue - an extra (duplicate-ish?) set of encounters.
OccurrenceID: BPC_OvrvwWtd_Fldr2_2019_11_03
Upload had the state set as “approved”. There were 26 rows (multiple animals in multiple images).

When I search under Approved Encounters, I see 26 encounters for this sighting.
When I search under Unapproved Encounters, I see 34 encounters for this sighting.

I’m going to check the encounters for the other 2 sightings that had this missing annotations issue to see if the problem is there as well.

Were there multiple dogs per image? I believe the 34 Encounters were created (by default in the “unapproved” state) as a result of finding multiple annotations per image.

Yes there were multiple dogs per image. So you’re saying that the 34 unapproved encounters are in addition to the original upload of 26 and created by the system finding the other dogs in these images? That makes sense. So all system generated encounters are set as unapproved. So then we’re good - all media assets in both approved and unapproved encounters have annotations - yay!


Now fingers crossed that we have found the culprit of WB-768…

Looks like it. Thanks!

Btw, any idea why that single encounter (https://africancarnivore.wildbook.org/encounters/encounter.jsp?number=dc8b03ed-ba88-416e-89e1-90e62b1ee0c1) in occurrenceID: BPC_OvrvwWtd_Fldr20_2019_12_05 did not have any annotations? It’s a virtual duplicate for one of the other encounters in the sighting (https://africancarnivore.wildbook.org/encounters/encounter.jsp?number=68109a7c-1272-4ae0-9f0a-68cdc6b7647e) which does have one annotation.
Is it just that the animals were not spotted in that particular image, as sometimes happens?