Zebra Summer Project Support Thread

A small separate issue noted in today’s meeting:

The new keyword-entry windows are working great in Safari on Mac. In Chrome on Mac, Maggie reports that every time she enters a keyword, the keyword-entry menu disappears, and she has to click out and reopen the window in order to enter another keyword. No big deal, she’s working in Safari now. But something’s up in Chrome.

Hey Andy,
Weirdly, we have confirmed that the delete is broken for the zebra platform (tracking with ticket WB-715). However, it isn’t anywhere else. That means we either had a small merge issue with the update or that there’s some custom code on the zebra platform that we have to track down and exterminate. We’re working on this right now; will let you know when there’s an answer (and a fix).

Regarding the Chrome issue: That’s incredibly weird as this was developed on Chrome. Finding someone to test this since I don’t have a mac and I have confirmed it’s not happening on Windows. Will get back to you with results. :+1:

Great and great. Thank you. If you need additional detail on the Chrome glitch let us know and we’ll get Maggie onto the thread to describe the problem she was having.

Hello, Andy!
I’m following up with Maggie’s glitch.
I wasn’t able to reproduce on my Mac (OS 10.15.15) and my Chrome (Version 84.0.4147.89 (Official Build) (64-bit)):

Thanks Mark. Since this is getting into granular discussion of how the glitch did or didn’t manifest, I’m asking Maggie to register, log in and respond directly. So hang on for that and thanks for looking into this!

  • Andy

Hi Mark,

I have Version 84.0.4147.89 (Official Build) (64-bit) of chrome and Mac (OS 10.14.6). I will try to update my Mac to see if it resolves the problem. I have attempted to attach a screen recording of the issue for more context. Another intern had the same issue. Thank you for the help.
screen-recording-chrome-issue-..

The red X delete button is working perfectly now! Thank you!!

Hey Tanya,
Today I have gotten this “error parsing results for task…” on 5 different searches today.

1.https://zebra.wildbook.org/encounters/encounter.jsp?number=f72019b0-6a80-4ce6-9896-f2f3c0dc6516
2. https://zebra.wildbook.org/encounters/encounter.jsp?number=7f0010f5-1c91-4c01-858a-9d939fde0957
3. Wildbook for Zebras
4. https://zebra.wildbook.org/encounters/encounter.jsp?number=bc72e969-febd-4651-b736-5de3e03635df
5. Wildbook for Zebras

Also, the ones I wrote to you about yesterday still produce the same error message.
Do you think the system is getting overloaded? Is there anything I can do to fix this error?
Thank you,
Eve

Hi!
I have been playing around with photos that a pretty zoomed out and contain several zebras. However, I have found that the algorithm will only recognize zebras in a zoomed photo that I have previously identified.

For example,
In this photo there are at lease 5 identifiable zebras,


https://zebra.wildbook.org/encounters/encounter.jsp?number=9b7b00d1-8527-43ec-862b-aca7f589e364
but the algorithm is only putting bounding box around zebras I have previously identified (which is two). But there are at least 3 other identifiable zebras in the picture that are not showing up in the matching results page
(Wildbook for Zebras).

Do you think that the other three zebras are not showing up because they are unidentifiable or is there something wrong with the system?
Thanks,
Eve

Hi Tanya,

For this particular photo, I have not been able to match. I haven’t had this issue for other pictures from the same encounter. For some reason the match results never loaded. Here’s the picture:

Here’s the link:
https://zebra.wildbook.org/iaResults.jsp?taskId=03891564-5f4f-4bdc-a69a-af042571adc2&nResults=4

Thanks for your help!
Safa

Hi Tanya,

I was wondering how to get the GPS data from encounters?

Thanks,
Maggie

The GPS data is available via the exif data in the photographs themselves. You guys have been set up with Alteryx Designer to pull that information out, so I’d reach out to @AndyGersick. If you need to get the information immediately, there are online options such as https://exifdata.com/. Upload your image and all exif data will be extracted. You’ll have to be careful of format though. I believe you’ll have to do a conversion to decimal latitude and longitude.

Tanya

I think the data these guys are having trouble accessing is just the result of their not having admin status and thus not being able to see all of the metadata entered by other interns (including location data). I can change their status as needed.

  • Andy

Separately I want to point back to the part of Eve’s post where she asked about cases where Wildbook is only catching previously identified zebras in group photos where there are other zebras that seem equally clear and indentifiable. Any ideas about what’s going on there?

It may be a role issue, but I’m not sure giving admin is needed. Let me do some digging and see if I can figure it out.

Regarding the data issues, I have an idea, but I’m trying to get confirmation asap. Please hold!

Alright, I’ve played with permissions around it and I’m not able to replicate the issue generically. Also, I think I need more definition on what the issue is.

@maggieliebich, can you give me the url of an encounter where you can’t see GPS data and a screenshot of the page so I know exactly what’s going on?

@AndyGersick,
We’ve dug into the data issues and turned up two bugs. First is that results are becoming available even though detection hasn’t completed (WB-734). Second is that, even though zebras are found and returned, an exception gets thrown, which kinda short circuits everything (WB-735).

We’ll pull these fixes in as quickly as we can. In the meantime, have everyone save any links to encounters where this is happening. Neither of these problems are happening consistently, so we’ll need help tracking them down as we go.

Hi, Maggie!
Oh weird. Thanks for including that gif.
Can I confirm that you’re still having this issue with our recent fixes?

@AndyGersick and everyone else!
We did some digging and figured out the issue with the encounters not displaying results (WB-734). It turns out the menu display was showing that results were available before detection was complete. Simply put, it tried to show results when there were no results to show. We’ve got it tracking through correctly now.
The exception issue (WB-735) has also been resolved. The long and short of it is that there were assumptions being made about two animals matching when they didn’t and the system didn’t know what to do with it. Good news is it’s fixed!

Thanks for your patience!