Are 'duplicates with multiple individuals in a photo duplicates?

Does that make sense. Is an apparent observation that has been duplicated, in fact not a duplicate if there are multiple individuals in the photo and the user makes no effort to note the records are for different ones?

Asking as I have had flags for duplicates removed by other curators removed for this reason, saying the presence of multiple individuals in the photo eliminates it as a duplicate.

1 Like

I would say those are not duplicates but ask the author to write description or find another photo if possible.

4 Likes

I strongly feel that the value of Inaturalist becomes severely diluted by numerous low resolution duplicates. What about a photo of a flock of 100 birds? Should this become 100 duplicates, even each with unique description?

3 Likes

Under site rules for observations, if you want to document that you saw 100 birds, that is exactly what you need to do, and is allowed by the rules.

Thus, none of these observations is a duplicate under site rules.

4 Likes

I’m not a curator, but I can absolutely see the logic in that:
Lets say you have a photo with 2 of the same species in it, one on the left and one on the right.
They aren’t duplicates, because you can say one observation is for the individual on the left and the cloned observation is for the one on the right.

It also fits because technically, each organism is supposed to be a separate observation (I’ve seen that explained in the old forum and elsewhere).

Personally, I am not going to separate something like these into multiple observations:

  1. https://www.inaturalist.org/observations/30650735
  2. https://www.inaturalist.org/observations/21878805
  3. https://www.inaturalist.org/observations/26141937
  4. https://www.inaturalist.org/observations/6057494
  5. https://www.inaturalist.org/observations/26082811

…but people who do separate them are following the rules much better than I am.

1 Like

i don’t think it is worth curators spending the time flagging them, unless the stated rules change. My guess is people who do this will probably tire of it pretty quickly unless there’s a really good reason for it, like particular annotations and fields associated with each given individual…

Either way, you can’t really use iNat data to determine abundance beyond a general overview, anyway. So it isn’t diluting the data either to add the six geese as six observations or one observation. If it’s birds you are interested in and you are interested in tracking abundance and absence, eBird is a better way to do that. Sadly there aren’t equivalents for all taxa (there is an eButterfly)

3 Likes

Assuming two observations with identical photo, having two organisms one at left and one at right, neither being clearly more significant than the other, and assuming there is no description or indication from the observer (such as an ID of one of them) to indicate which is the subject of the observations:

Encountering the first observation and being the first to ID, I would ID as the one on the left. I would then leave a comment that “I am assuming the subject to be the one on the left, please let me know if this is not so”

Encountering the second, and again being the first to ID and assuming I remember having seen the other observation, I would go back to that other observation to compare to this one, and seeing that they are the same photo but two possibilities for subject, I would ID as the one on the right, leaving a similar message as to the first.

If, on the otherhand, one of the organisms is slightly more “significant” in the photo, then as above, but substitute “most significant” for left, and “second most significant” for right.

5 Likes

I don’t think they’re worth flagging.

Does this happen often? The only time I’ve heard about it becoming an issue was when a user in Canada was doing this for flocks of birds because they wanted to win the camera for the Canada Bioblitz a year or two ago. Even then it technically wasn’t wrong, it just wasn’t a great use of iNat, in my opinion.

5 Likes

image|666x500

This is an example of how I do it. I circle the individual I’m identifying and then do separate posts. I’ve gotten comments from people demanding that I consolidate them into one observation, but I’m not sure why that bothers anyone. They’re individual organisms and individual observations and I think that’s worth recording separately.

4 Likes

The downside of this, is that any image that is similar to yours is not going to correlate in the AI/CV as similar, because of that added red ring. Or worse yet, images with similar red rings might correlate, an contribute to the errors in CV suggestions. It’s better from a logical perspective to provide clarity in the description as to which is the subject (in this case “left of centre” would suffice), or perhaps put the modified image into the description or a comment with an image tag, and then delete the modified image from the observation photos (iNat still retain the photo you upload, even after it is deleted, so you can link to it in your image tag)

1 Like

If you’re interested in logging two observations for two individuals, that’s totally fine. If people are “demanding” you combine them, you can point them toward the FAQ: https://www.inaturalist.org/pages/help#observations1

You don’t need to necessarily circle it in the image, but I also wouldn’t really worry about what the computer vision model is training on. It has plenty of photos of raccoons.

4 Likes

I believe that’s the opposite of how AI learning works, any images that have distracting things like red rings help the AI learn to recognize a raccoon, because only some of the images will have a red ring. Just like only some will have letters from a trail cam, some will be at night, some will be during the day, some will be blurry, some will be close-ups. The only thing all the images have in common is the raccoon, and that’s how the AI learns what a raccoon looks like, and what it doesn’t look like. So to get the AI as good as it can get it’s beneficial to upload loads of crappy images with distracting visual elements.

6 Likes