Encounters have no images

What Wildbook are you working in? grouperspotter

What is the entire URL out of the browser, exactly where the error occurred?
https://www.grouperspotter.org/encounters/encounter.jsp?number=cce8d815-a6d4-49de-81a7-21fb9b00003c
https://www.grouperspotter.org/encounters/encounter.jsp?number=cce8d815-a6d4-49de-81a7-21fb9b00003c
https://www.grouperspotter.org/encounters/encounter.jsp?number=076ef914-d38f-43ca-8373-c5a32ddea054

Can you describe what the issue is you’re experiencing?
Encounters uploaded on 6/27/2021 do not have images connected to them, although there may be more like this from different dates. I’m certain all the encounters should have at least one image. There is no error message, just an empty space in the Gallery section.

Can you provide steps on how to reproduce what you’re experiencing?
Looked at encounter, and realized there were no images attached to it.

If this is a bulk import report, send the spreadsheet to services@wildme.org with the email subject line matching your bug report

Hi @BTran

We’ll need more info to diagnose these.

Did you bulk upload these?

Did the bulk uploader report any files missing?

Do you know which images you expected on these encounters?

Thanks,
Jason

Hi @BTran

I did some more investigation, and these empty Encounters are an artifact from when a single photo contains multiple individuals (specifically >=2 annotations are found by the detector) and therefore a new Encounter was cloned (because an Encounter can only be one animal, and two in a photo means there are at least two and a new Encounter is needed).

When we reset your data to resubmit for reprocessing, these Encounters were left behind. They can be safely deleted, or if you want, I can remove them in bulk.

Thanks,
Jason

1 Like

Hi @jason,

Thank you for the update. We spent a lot of time last night getting frustrated over this and trying to uncover why the encounters were that way.

Happy to hear that there is a simpler solution to things.

Let me discuss some stuff with the project managers and I’ll get back to you on what to do,

Thanks,
Bryant

My project group also had some other questions they’d like to ask:

Where do we enter info about being bicolor?
Bicolor is a type of color pattern for nassau grouper, and we’re wondering if there is an option for this.

If we name two different sides of a fish the same ID will the algorithm think we are confirming those two sides as a match? How does it associate them together without “matching them”?

Where can we add information for an image like “Laser present” or “Tag Present” and Tag Color?
We have tagging information of the fish that we never added to the bulk imports. Would this fit on any column in the excel sheets sent to bulk import?

Thank you,
Bryant

Where do we enter info about being bicolor?
Bicolor is a type of color pattern for nassau grouper, and we’re wondering if there is an option for this.

This sounds like a good use of Keywords, which can be added to each photo on the Encounter page. There’s a bit of documentation about it here:

https://docs.wildme.org/docs/researchers/encounter_guide#keywords

I went ahead and added “Bicolor” as a supported Keyword here:

and then it will appear as an an option when you zoom into a photo on the Keyword page:

Individual IDs are set at the Encounter level, and an Encounter is all the data about one animal at a point in time. All annotations (the green bounding boxes in the Encounter photos) found for the photos of each Encounter are assumed to belong to the same animal. All Encounters with the same individual ID are assumed to also be the same animal.

These would also be a great use of keywords, which can be bulk imported. The relevant column is Encounter.mediaAssetX.keywords.

1 Like

A post was split to a new topic: Bicolor ML detection for groupers

All of these keywords would go in one column?

So if we are looking at an image in an encounter that has multiple annotations. Those additional annotations have been assigned a new clone encounter? Should we keep them then if the image is good? We were a little confused as to what to do with all of those separate annotations on one imate.

@jason for the column names would we be limited to the preset column labels for the metadata, in the excel sheets?

Can we create our own column labels, if we want something to be labeled in a unique way?

We’re asking this because, if we have 20 images of the same individual, and they were all labeled with mediaasset1,2,3, etc…, but only one of them had a tag in the photo for it’s right side, but the others are from the left side, without the tag visible, how would we specify this during the bulk upload in the excel sheets, if the keyword would be in the same row as the row with multiple image file names?

That is correct. They go into the one column. There is an an alternative way of specifying them, but it assumes that only one keyword per photo. Encounter.mediaAssetX.keywords allows for any number of keywords per photo.

By making sure the Encounter.mediaAssetX.keywords column has the same X value as the Encounter.mediaAssetX column. Example:

  • Encounter.mediaAsset0 = IMG_0123.jpg
  • Encounter.mediaAsset0.keywords = tag

Do you mean your own column heading titles, such as Encounter.myVariable? No. We would have no way of understanding what a random column name would be intended for or where to put it.

Correct because an Encounter can/should only represent all the data (visual and otherwise) about one individual at a point in time.

Should we keep them then if the image is good? We were a little confused as to what to do with all of those separate annotations on one image.

Yes, this is data. You have a mark-recapture data point for another individual.