Incorrect IDs added automatically when duplicating an existing observation

There seems to be a new difficulty during the process of duplicating observations. I posted an observation of a Western Honey Bee on a White Melilotus. Then I duplicated the observation so that I would be able to use it for an observation of the White Melilotus. By the time I could see the new flower observation, it was automatically identified as a Spider Orchid. I added four new photos of the flower, deleted the cropped bee photo I started with, and changed the ID to White Melilotus, and the result is an observation showing that I disagree with myself for originally identifying the flower as a Spider Orchid (which I never did). Is this how this is supposed to work?

If a Computer Vision ID is necessary in the making of a duplicate observation, could there be a lag time before it’s made? If not, could the Computer Vision icon appear at least to indicate that the Spider Orchid ID may not have been intentional?

Oh, this might not be a reproducible error. You can edit and delete your IDs, if necessary.

I just now duplicated the same cropped bee photo. I didn’t add or delete any photos this time. The Computer Vision made an unsolicited ID of Spider Orchid without any participation by me, as before, but this time the Computer Vision ID disappeared after my correction. So I guess it is not a lasting problem with duplicating observations. But it seems new to me that the Computer Vision has gone beyond offering suggestions to actually adding IDs without human participation. And the lack of the Computer Vision icon on such IDs seems new, too. But I think what bothers me most is the possibility that Computer Vision IDs may be attributed to a person who did nothing to select that ID, instead of to the Computer Vision.

3 Likes

Don’t you have to accept the CV suggestion for it to be assigned?

I’ve never had it assign an ID without me accepting it first. Sometimes I’ll hit the accept button by accident if I’m in a hurry, but that’s always my fault, not the system doing it automatically.

That was my understanding before this happened today.

I don’t think I could have duplicated it a few hours later if the first time was just an accident.

1 Like

I just uploaded the same photo. The Computer Vision identified it as Spider Orchid again. I did not agree or do anything but upload the observation here: https://www.inaturalist.org/observations/36993988

1 Like

Just now I made this observation: https://www.inaturalist.org/observations/36994106. The only things I did were hit Upload, hit Choose for the photo, and hit submit for the observation. Again, the Computer Vision made the ID, but attributed it to me without my participation as to the ID.

2 Likes

I’m adding screenshots for the last observation I linked, in case it changes:

1 Like

I’d guess your problems aren’t related to computer vision at all. There is some new feature in place that auto-assigns taxa based on file names. Problems around this have been reported before. Could it be that you are being hit by the same?

6 Likes

I don’t consciously do anything with file names. I don’t even know how to do that. I take photos with my camera, load the photos onto my laptop, and without doing anything further to them, I upload them to iNat.

This is the other thread I was referring to: https://forum.inaturalist.org/t/add-observation-automatically-tagged-my-observations/8203/19. It has happened to others with simply the default file names as they came from the camera or phone (for example caused be months names like ‘May’ in there).

Don’t give up on iNat yet. It is a very friendly and helpful bunch here and I’m sure your problems will get sorted and if it’s a software problem it will get fixed. This time of the year is probably the worst though for getting quick responses from staff/mods. They deserve some time off too sometimes!

5 Likes

@sgene Do your photo filenames happen to have “ada”, “bd”, or “bsal” written anywhere in them? Those are some of the names on those taxa:
https://www.inaturalist.org/taxa/140934-Brassia
https://www.inaturalist.org/taxa/948980-Batrachochytrium-salamandrivorans

2 Likes

Where are file names? How do I check them? I don’t know anything about them.

1 Like

The file name is the text that appears on the photo before the “.jpg” part of it. EG for these it’s “DSC02966” and the like:

1 Like

Yeah, I’m not sure what the photo file name is when you’re duplicating an existing observation. I would think it would just be “original.jpeg”, from https://static.inaturalist.org/photos/58570626/original.jpeg?1577436401, from right click>Copy image address here: https://www.inaturalist.org/photos/58570626?size=original, but I’m not sure what is going on on the back end.

When I duplicate an observation, nothing ever appears in the ID section:

image

image

1 Like

Thanks. I’ve never seen these DSC numbers or .jpg numbers at any other time from taking the photos to uploading them to iNat. I may try to find them later, but I really don’t even like the idea of having to get involved in these kinds of things. At least I understand generally what has happened now.

It’s okay with me anytime you want to close this topic or change the title, since I was off base as to where the problem was. I went ahead and fixed the Melilotus observation and deleted the duplicated observations I made just for this topic.

1 Like

I don’t see any of those exactly, but the words “Add a” appear several times.

1 Like

On the one used for the bee and Melilotus observations I just get the same “Add a” but no “bd” or “bsal”, (and I didn’t get a map this time because it’s a different camera with no GPS, unlike the stork photo taken with a camera with GPS, if that makes a difference to the ID problem):

@bouteloua, you asked me about “ada”–what do those letters relate to?

It was listed as an unaccepted synonym for Brassia. I’m a bit stumped on this one myself.

1 Like