We might also consider the license terms chosen by the user, for the photos.
But I understand that this is not a matter of rights, it is a policy.
iNat’s choice is to avoid making the observer uncomfortable, in no way.
So, a button for the observer only (for fixing) and/or a DQA flag for all of us (for filtering).
2 Likes
2 Likes
Automating or somehow streamlining the process would help at least a little, I think. I wrote a tutorial on how to do it manually, which I link on observations I come across, and sometimes people reply that they tried but it was too confusing.
2 Likes
I used to think it would be good if identifiers could split up multiples.
But, we don’t know where each one was. Say the pictures were taken along a hike.
Only the observer will know where each one was (my pictures don’t have GPS attached)
Multiples need to lurk in Casual (as in lacking data - Not Cultivated / Captive), waiting for the observer, one day.
Meanwhile the example I offered is no longer multiple - but the observer has opted out of Community ID so it will always be wrongly IDed.
2 Likes
Moreover, if no GPS data is attached, impossible to automate the split (that would be based on computer vision, for keeping together photos of the same species) by just pressing a button, even for the observer. In this case, as you say, only the observer can split, and only manually.
The worst case would be an observation with GPS data attached to the 1st photo, but not to the other photos. The observer is not notified at upload time (location known) and then he has time to forget about the other locations before someone notifies it is a multiple species observation. Then, the split is impossible (other locations unknown).
3 Likes
Nice! I could probably add 1-2 K obs to that if I went back through my prior interactions. But quality would be variable.
It’s great to have the different-language guidance links right there on the project page. My hope is that someone might make a Chinese-language(s) one someday. Russian might help too. Others on anyone else’s wishlist?
I like to show the community of plants that the observed plant exists within as the first photo in the observation. Subsequent photos show the intricate detail of the plant that the observation is for. I do this because I’m trying to do an ecological restoration of my backyard. It’s important to me personally, to show colony expansion / contraction within a community of plants but also hi-light the species. Is it ok to put the “community” photo first? I like putting it first because I want the community photo to show up on the “stats” page showcasing my restoration. Brand new to the forum so don’t hesitate to tell me if I’m doin’ it wrong.
Example:
https://www.inaturalist.org/observations/73757618
As an aside… I also wish my observation “notes” appeared as an “onMouseover” event when the cursor is hovering over a tile on the “Stats” page… it’d make the “Stats” page similar to a blog of observation notes having the most Comments and Faves. The “Stats” page for umbrella projects would become similar to a blog with multiple “authors”.
1 Like
You can use any photo you wish in any order as long as object of observation can be seen (often people use photos of a habitat where the organism is not present, that is a thing to avoid). I wouldn’t even call your example a question inducing, it’s a totally normal photograph of group of plants, that you shouldn’t worry a second about.
4 Likes
I guess I was more interested in how people think about it. It seems like iNat was designed to be granular… at the species observation level. I always think I’m using it in a way that it wasn’t initially intended. In a breadth of knowledge (ecological) way rather than a depth of knowledge (species) type of way. I love the way in which iNat allows you to combine the two and hope they keep moving it in that direction. That’s where my “aside” comes in… a way to add breadth across granular observations within child projects which are themselves inside an umbrella.
1 Like
Welcome to the Forum!!
Parallel situations? If we want to observe a House Finch and we take a photo of it with Blue Jays, we won’t necessarily crop out the Blue Jays. We will just ID it as House Finch and since House Finch is present, it will get ID’d as such despite Blue Jays.
Further, if we record audio of birds, many species may be vocalizing and so we will upload the observation with multiple species “present” and tell through notes and/or our initial ID what we want identified. It seems to always get ID’d as we intend and no one criticizes us, though they might suggest, “You could duplicate this observation for the Blue Jays.”
3 Likes
It’s easy to miss 2nd and next photos are different, or it can be 1 of 7 photo that is of a different organism, ider sees what they can id from the first pic and just doesn’t click on the next.
1 Like
I ID multiples back to Life, if someone has chosen to ID a random photo. Confusing the taxon photos is spreading the problem beyond the initial obs.
4 Likes
I understand and like what you are doing, but consider that you might not get an ID as quickly by doing so. The computer vision looks only at the first photo - it works better on photos that are cropped closely to the subject. There is also the problem that if you are observing a species that is not common enough to have been used in the training set for the computer vision, then your community photo might end up being used in the eventual training set for a model that might then be less precise. I’ve seen the CV identify plants as beetles and birds as plants for that reason.
A slightly different story applies to the “Identify browser.” There you’ll see a thumbnail of your photo(s), and if it looks too challenging (messy), the potential identifier with hundreds of observations to identify might choose to skip yours and focus on the easy ones. Finally, to identify plants to species commonly takes some very small characters such as whether leaves have folded margins, twigs have hairs, or sepals are reflexed. All of which are probably impossible to see in a community photo.
One tactic you could use to speed up getting IDs would be to put a photo with the correct details in the first (0) position until the observation reaches Research Grade. Then you could reorder your photos so that the community photo is first.
6 Likes
Seconding that. If @stockslager hopes to get an ID for their obs, start with the cropped photo showing THE distinctive feature, followed by any other distinctive features, and end with the wide view for scale, growth habit, habitat. All useful info but identifiers are swamped in Unknown and Needs ID.
4 Likes
My own concern with this issue is that I would like not to see anymore unrelated pictures in a taxon page.
This makes sense, but seems like it could easily be handled by suppressing anything tagged multispecies from appearing in taxon pages.
More generally, multiple-species observations degrade the quality of iNaturalist contents. …
Another concern is the loss of potential value of these observations. …
I think this view implies that iNat’s mission is perfect taxonomy and ID of individual organisms. Maybe it is, and I certainly find it useful, but that seems like a huge missed opportunity for doing systems science. Knowing that 5 different insects are feeding on the same flower at the same time from a multi-obs is far more information than what you could ascertain from the geographical and temporal correlation of the same entities in individual observations. Reduction to single obs strikes me as the real loss of value.
In my own usage, as a relative noob, I know that there are good data points that I avoid submitting because it would be a hassle, or seem unnatural, to duplicate the item and point out different features. Songbirds bothering owls come to mind, but I’m sure there are many relationships that would be useful if the data captured them.
2 Likes
After reading most of this thread, it seems to me that many of the challenges expressed go away if this is approached as a user empowerment problem rather than a moderation problem. As long as there is no mechanism for recording a multispecies observation, you’re stuck relegating them to some kind of purgatory, which immediately raises concerns about editing user data and whose liver is it, anyway?
The obvious solution to me (admittedly as a relative noob) is to create a way to mark and link multiple species. Then users (the majority of whom I would presume are conscientious) would solve most of the problems bottom-up. If there were a way to split exiting observations, it would be a simple matter to give a gentle poke indicating that users need to do so (and reasonable to leave them in purgatory if they don’t).
3 Likes
First off, welcome to the forum, @tomfido!
I don’t think this is the kind of multiple-species observation that Jean Philippe is talking about. If you have a photo of five different insects on a flower, there is no problem at all duplicating that observation and identifying each species present, although it can be helpful to crop it and put a cropped photo of each species as the first photo of each relevant observation in turn. You can also add observation fields to make explicit the interactions, such as flower visitors. That would be the best approach, and is to be encouraged for examples like:
The sorts of multiple species observations I think Jean Philippe is talking about are usually from new users, who have gone out and taken a photo of, say, a daisy, an oak, a squirrel and a beetle, and upload them all as a single observation. Those need to be split into different observations to be useful and identifiable.
6 Likes
In that case, I withdraw my comment, and will look into the use of obs fields for recording interactions.
1 Like
I have an update to this request, using a project as a workaround:
https://www.inaturalist.org/projects/variable-species-and-other-life
1 Like
After looking at the options, I think they fall a bit short of what’s needed.
For the problematic hawk-bee-poodle combo, the kind of user who creates that probably needs a simple way of selecting one or more mismatched images, and saying “split these to a separate observation” in one go.
I’ll use the interactions observation fields in the future, but that’s also discarding some information. The interactions tell you harassment/predation/etc. are going on, but not which other obs or species is involved. Interactions>Visited Flower Of… is a little better, but still links a species rather than an observation. Seems like what you really want is an obs x obs table with a link property.
2 Likes