Bulk import giraffe spotter
We are two students from Copenhagen University who have been in South africa to collect data. This data collection has been ongoing for several years. Before we import our data (2023), we wish to teach/tell the AI which individuals are who, from the previous years. Do we need to import these for AI detection? Or is it enough with the regular import assigning individual names?
Another question, is it possible to cancel/stop an import sent to AI detection?
We have previously imported big bulk imports, that are too big, that we would like to cancel, so we can send the smaller imports.
I have attached the link to those we want to stop the AI
Best Holiday wishes and kind regards,
Nadja & Laura
Hiya @lpurlund ,
To start, I’ve gone through and deleted the bulk imports you listed here, so those should no longer be blocking you
Before we import our data (2023), we wish to teach/tell the AI which individuals are who, from the previous years. Do we need to import these for AI detection? Or is it enough with the regular import assigning individual names?
Use the regular import with assigned individual names. That is all we need for detection and ID suggestions to populate. And at a later time, if we do model retraining for these species, that assignment will be recognized and make the model better.
Hope that helps! Happy holidays
Thank you so much for your help. All the ‘old’ data with the assigned individuals have now been uploaded.
Our first ‘batch’ of 2023 data (with no assigned individuals), is up for detection. Does it look correct? And how long will it take you think?
Hey @lpurlund ,
Happy to help. Sorry I missed this before, but I did some spot-checking of your encounters it looks like your imports are missing the species being set. Wildbook will not run detection successfully without a species.
For each encounter in your bulk imports, you’ll need to include the following two columns:
And then re-upload. I’m so sorry about that!
All the best,
Thank you for the reply!
Is it only the 2023 data (the data sent for AI detection), or also the “old data” with the individual assigned?
Hi there, jumping in here. The species information is required on all encounter records in the system in order for them to be able to be processed by the Wildbook Image Analysis - detection + identification.
Thank you for your help and reply. I think we fixed it. Does the detection look correct?
Hi @lpurlund - I don’t have access to GiraffeSpotter, so I can’t look at your data. I manage ACW, Whiskerbook and Wild North only.
However, if you can see annotations around each giraffe in each photo, then detection has been successful.
The first bulk import is still just loading, as if it is stuck.
@tanyastere could you maybe take a look and tell us what we are missing?
Thank you again!
Can you email the spreadsheet for this bulk import to email@example.com so I can take a look at what might be causing detection to get stuck?
Thanks for sending your spreadsheet!
Encounter.specificEpithet fields are case-sensitive. You have both entries in capital letters but
Encounter.genus needs to always start with a capital letter and
Encounter.specificEpithet must always start with a lowercase letter.
You’ve entered “Giraffa Giraffa” so you’ll need to make sure it’s written as “Giraffa giraffa”. When WIldbook doesn’t recognize the species as entered, it can’t begin detection.
You’ll want to update your spreadsheet, delete your import, and re-import the new spreadsheet.
Make sure to review our Bulk Import Cheat Sheet, our Bulk import help docs, and our Bulk Import training video series to avoid the most common issues with imports.
Thank you for the quick help! It’s been up for detection for almost a week now (with the newest edits), is this common to wait this long or could there be a different issue?
No, it shouldn’t ever take that long to complete detection. If you notice an import stuck in detection for more then a few hours, let us know so we can dig into the issue. Can you share the link to the stuck import so I can research this further?
I think I found it: Wildbook for Giraffe | Login
I was able to manually force it through detection and it’s ready for ID now.
Amazing! Thank you Anastasia
We have now sent it to ID
Unfortuantly it is assigning our 2023 individuals with our other 2023 data, and not our previous old data (which aldready has assigned specifik individuals).
As mentioned above and in our e-mail thread, we have recieved old data from 2012 to 2021. These encounters have already assigened individuals, which we would like to ‘teach’ the AI with (this is oploaded, but not sent to detection as you adviced). So when we import our 2023 the AI photo identification will tell us who’s who.
Have we done it corretly or has there been some miscommuntiocation?
Also, we notticed it giving and error message regarding the photo identification.
Again, we really appriciate all the help so far and our very thankful for you time!
Nadja & Laura
Please see this morning’s post about the server outage for updates: Outage: January 30, 2024
The outage has been resolved. Let me know if you’re still unable to work on matches.
Still the same as commentated earlier. Our 2023 data is matching with our 2023 data and not our so called ‘old’ data, with the given individuals.
Regarding the “Something went wrong with fetching the inspection image” error, that just means it couldn’t load the Hotspotter heat map. It appears to be working now:
I compared a few encounters in this import to your import for
Wildbook2012.xlsx and noticed that the Location IDs are entered differently. The 2023 import link you shared has the location ID as “Pilanesberg
: South Africa” and the 2012 import has the Location ID as “Pilanesberg National park
: South Africa”. If your locations are not entered exactly the same, GiraffeSpotter will treat them as unique locations and will not compare the giraffes seen in one against the other.
It looks like on May 9, 2023, I sent you an email asking you to post a feature request in Community for it to be added to GiraffeSpotter. Because the post wasn’t made, the work never started on it. I’ll make a post to track it now so it can be added soon: New GiraffeSpotter Location ID : Pilanesberg National Park
Are you including your user ID in your bulk import spreadsheets? When I searched GiraffeSpotter for Encounters assigned to your username, I got 0 results. You should see your name on the Encounter’s metadata when it’s assigned to you:
Not having your encounters assigned to you means that you won’t be able to confirm matches on those encounters. We can fix this retroactively so you don’t have to delete the imports that are already uploaded, but it’s something you need to fix going forward. You have to include
Encounter.submitterID as a field in your spreadsheet with your GiraffeSpotter username in it for each Encounter on there.
I’ll let you know as soon as I have the new location ID in GiraffeSpotter. Then I will work with my team to assign all of your bulk imports to you. ID will likely need to be re-run so that the results all display giraffes from one, single location. I need to find out if we need to handle this step due to the volume of encounters you’ve uploaded in order to not strain the queue for other users.