Add a "visually identifiable" scale parameter to set user expectations on confidence of computer vision IDs

oh right! not sure i knew that existed
useful
i’d love to have a quick way to add those
a hotkey even

1 Like

You could probably create your own macros.

I saved the link to responses to a journal post, but I usually end up googling “inat frequent responses”, because that takes me just as few clicks as visiting my list of iNat bookmarks.

1 Like

is there a list also somewhere of the existing hotkeys for identifiers?

[EDIT] - found another feature request explaining the hot-keys
https://forum.inaturalist.org/t/add-10-new-hot-keys-to-the-identify-page/3122

1 Like

I prefer to engage with new people by writing my own comment.

Out of the box copypasta comments - I dunno. Most of the professional scientists have their own set of comments to fit their own cases. I started to recognise and appreciate the … can’t ID Bougainvillea to species from this photo …

1 Like

Sure…and I wouldn’t want to disable those who want to go the extra mile! I think enabling some people to have a short-cut copy pasted option if they want it though, would be great ( and could enable data to be harvested relating to it )

It shouldn’t be at the cost of people who prefer to write something personal each time though :slightly_smiling_face:

2 Likes

Hello

This has not seen activity in a while but I think this is still an important topic. I agree that this would be a lot of work but focusing on the species groups that are most likely being confused would still be a big help for citizen science in learning how the distinguish species and/or learn that some are not distinguishable by morphology alone.

I have an example page that use a 3 level scale for distinguishability.
The German site kerbtier.de is focused on beetles and you can very quickly see whether a species is distinguishable by morphology. By looking at the green/yellow/red indicator.

This would be a great first step.

A further step would be displaying a notification when people try to select a non-/hard to distinguish species and recommend staying on the genus level unless they are sure.

The last level of awesome would be if iNaturalist would give recommendations for how to distinguish hard to identify species. maybe including hints to use other clues than just visual appearance. For example: “This species cannot be distinguished by appearance but here is it’s distinct call!” “These species look the same. They can only be distinguished by dissecting and observing their reproductive organs. We recommend staying at the genus level”

Thanks for surfacing this old thread. I thought I’d commented on something similar somewhere but I’m not sure. I think something like this would indeed be valuable, and in response to concerns expressed above that it is a huge undertaking, I think it would be sufficient to automate it based on iNat identifications. We already have the ‘similar taxa’ functionality on the taxon page which is based on how many observations have disagreements. I’m sure it would be possible to judge the red/yellow/green system based on the number of, (or better - the proportion of) observations, in the taxon subject to disagreements/bump backs etc.

1 Like

here is a really rough visual prototype for iNaturaslist web.

I’m not a UI designer and it’s the first time I used that tool. So I couldn’t get the hover function to work correctly. You’ll have to click somewhere to close the popup.

I hope this at least gets the idea across

Your particular example is an interesting one because it raises the question of how geography affects difficulty. You’ve given Chorthippus brunneus a red. It’s very difficult to identify this species in continental Europe - but very easy to identify it in the UK where the similar species are absent. A more sophisticated algorithm for the colours might perhaps be able to use the geomodel to see if the range overlaps with the ‘similar taxa’, or perhaps might only use the proportion of disagreements within a geographical range.

3 Likes