Let's Talk Annotations

Is it possible to change or delete annotations on observations that aren’t your own? (I only see the “approve”/“don’t approve” buttons, but no way to suggest a different value…)

If it were possible, we could just have all of them at once. Then the observer and identifiers could decide on a case-to-case basis whether to apply “female” or “worker”/“queen”.
(If it isn’t possible, I think it probably shouldn’t be implemented this way to avoid confusion)

It is not possible to change/delete annotations on observations that aren’t your own, only to up and downvote. You can see some discussions of this elsewhere on the forum.

Would you be comfortable with the breakdown “female” / “worker” / “male” as options under the “sex” annotation for all of Vespinae? I want to save the iNat staff the extra work it would take to add different annotation options for every species.

Agreed, there should be a distinction between “ripe” and “unripe/developing” fruits (though obviously both can be present at the same time).

What you claimed didn’t seem quite right to me, so I did some more research. The term “queen” is frequently used by Vespid researchers to refer to female obligate social parasites throughout the literature, most notably in Akre R.D. et al. (1981), which served as the definitive text on American Vespinae for several decades.

It makes sense to use the word “queen” in this context, because even though obligate social parasites do not produce workers themselves, they adopt the role as the dominant reproductive entity in the colony. They are able to mate and the workers of the colony are not.

1 Like

I don’t think an extra annotation value should be added (for obvious reasons I hope!), but here is a type of evidence I have not yet seen and does not fit any of the current categories - the defensive fluid produced by a mole cricket

Heads up, we’ve just released some updates to annotations for angiosperms and vascular plants. Many thanks to everyone who left comments in this thread that informed the updates.

6 Likes

I just wanted to say that I’ve already passed 2,000 fragments! https://www.inaturalist.org/projects/seashell-fragments
To be fair, I’ve also been including crustacean and echinoderm fragments, however, I am also acutely aware that they are all concentrated on the US Atlantic and Gulf Coast which is the only place I have expertise in. Only a couple hundred are in Europe and other areas which means I have a lot of work to do!

2 Likes

Vegetative growth is an important phenology stage, please to narrow to just reproductive phenology annotations

1 Like

Sorry if this has already been mentioned (I confess I haven’t read the whole of this very long thread).
For a few days now, the plant annotations have not been translated into French :
image
Is it possible to fix this? (yes, it’s true, the French aren’t very good at foreign languages: thank you for them/me!)

I’m happy to contribute to the translation if someone tells me where to do it :smiley:

3 Likes

yikes, 670 posts…!

I would like to see a tool for highlighting identifying features, not just for human’s identification, but also to help direct the algorithm.

so for example plants have stems that are smooth, stems that have a polygonal profile, stems that are hairy; leaf margins that are smooth, lobed, toothed… etc

if you’ve ever submitted feedback to Google about an app, then you’ve probably experienced an option to upload a screenshot and given a chance to highlight what specifically you want the Devs to look at. a similar method was used by a citizen science project that I participated in, where the idea was that the citizen scientist would look at images and then highlight the items of interest the project was looking for.

so in addition to adding drop-down options to choose from these common identifying characteristics, there should be a tool to use this highlighting layer to highlight exactly where on the photo this identifying characteristic is. and FYI the Wikipedia entry excerpts are pretty useless for identification. sure I could do a Google search about the differences between two similar species, to confirm which variety I’m looking at, but when there’s 16 different options that becomes pretty tedious, whereas if I’m using the “compare” feature, it would go a lot faster if I knew what features to look for because it would highlight them on the image for example for sassafras there are two options, wild sarsaparilla and bristly sarsaparilla… but, not being familiar, and only having one example of the two in front of me, I’m kind of left to guess about the “bristly” adjective… nothing about the plant I’m looking at looks particularly bristly, but maybe refers to something more subtle like the flowers or the appearance of the leaf margins both of which are toothed.

also having this feature would incentivized taking photos of these features, because you can’t highlight it if you don’t take a photo of it. this would also help the algorithm on some of the blurrier photos. I remember suggesting to someone that they upload non blurry photos, and they replied “you can clearly tell because of the white patches on the shoulders” that doesn’t help the algorithm though, and it doesn’t help anybody who doesn’t already know that white patches on the shoulders are the differentiator… all I could see was a bird-shaped blob against the blue sky.

edit: I should also point out (though I probably don’t need to) that most people are not going to go to the effort of highlighting all of the characteristics, instead they’re going to shortcut by only listing the most differentiating.

The CV algorithm doesn’t learn features; it compares photos with the labelled photos in its training set. The training doesn’t distinguish between blurry photos and good ones. It is actually reasonably good at recognizing blurry bird-shaped blobs because there are plenty of RG observations of blurry photos that skilled birders were able to identify (based on their knowledge of features like white patches on the shoulders).

I do think that being able to annotate photos might be useful, but I’m not sure it would have any direct effect on the CV; I suspect the computer learning model would have to be completely reworked to make any use of such annotations.

For what it’s worth, I find that field guides (online or paper) and similar resources are a better way to learn about identifying characteristics than comparing photos. The point of a field guide is to help you sort through the various possibilities. As you note, there are many taxa where the visual differences are not intuitive. And particularly for difficult and underobserved taxa, not all species have observations on iNat, so it often isn’t a good idea to rely solely on the “compare” feature to determine what species are possible.

There have been some discussions about how to integrate identification information and resources on the taxon pages, but there are non-trivial challenges connected with this.

1 Like

yeah I just found one from a year ago where the OP described marking XY coordinates of a fruit. basically the same idea, and the reply was that a solution would need the annotations to be “per photo”, and while that is true, we’re talking about multiple separate components here:

and, if they did it the usual way, the database is completely separate from the rest.

if true they can add as many fields as they want to the database and the program will ignore them until you give it a reason to use them. the tool could describe the polygon coordinates of the highlighted area, referencing the specific photo, and then save those coordinates to the database. with a separate version of the program you could pre-populate a bunch of these, then, when the new version of the program is ready with the necessary UI changes, those fields will already be in the database to be used. that’s a simplification of course, but it means that they could add the fields now for the identification characteristic annotations
without even worrying about any kind of markup method.

never mind identification, purely from a data standpoint, that would enable somebody to search the data for “all plants with waxy leaves in a certain area”

I am not sure if this has been brought up before, but has there been talk about potentially adding shortcuts for annotations? Especially when going through in the identify tab, I find it quite cumbersome and slow to switch to the annotation tab to input the sex, so I normally don’t and just include it in the comments alongside the ID. I know that this might be tricky with taxa that don’t follow the strict male/female category, though.

1 Like

There already are some existing shortcuts for annotations. You can see an example workflow here:
https://forum.inaturalist.org/t/interpret-annotation-shortcut-as-upvote-when-it-duplicates-existing-annotation/24423

There are also some customizable, user-made extensions that can do more complex workflows, eg:
https://forum.inaturalist.org/t/announcing-the-universal-metadata-tool-beta/53182
that you could look into.

5 Likes

What is the apt category for
Mantis ootheca (construction or egg)
Spitttle bug spittle secretions or spittle bug larval foamy or solid ‘case’

As in adding shortcuts to annotate while you’re on the Info tab? Or when you’re on the Annotations tab? Note that you can use keyboard shortcuts to go between tabs in the Identify pop-up: SHIFT + right or left arrow.

1 Like

for ootheca I use Construction and Egg

Seems reasonable. It is basically a nest with eggs inside it I think? Could do the same for an unhatched spider sac.

1 Like