Using Computer Vision Symbol in Deciding Subject of Observation

Here is a hypothetical example:

An observation has a photo of a plant with an animal. The observer’s ID is of a plant and has the computer vision symbol next to it. Subsequently, other people add identifications of the animal, citing “computer vision error.” The observer doesn’t respond.

I have seen a number of observations like this in the State of Matter Life category. My inclination is to assume that even though the observer’s ID may be incorrect as to species, the subject of the observation is an organism in the photo that is in the same category as the observer’s ID. I’m interested in the views of other people with respect to adding identifications in a different category on the basis of the computer vision symbol.

2 Likes

I would generally agree with you here, although admittedly I have seen a few weird cases where the opposite was true, where the observer ended up responding and saying they actually wanted eg the bug on the flower instead of the flower itself, despite having picked a CV suggestion for a plant species (so they clearly just clicked any random CV suggestion without checking whatsoever)

4 Likes

In a subset of those, I’ve seen that the observer has reported difficulty changing over their id in whatever interface they are using, typically as a newish user, so they appreciate the “counterweighting”.

Anyway, in those “I do think they really meant the spider not the plant” type situations, I might offer the spider id, with this comment:

“if you want the plant instead (which you chose from autoid [remark added if applicable]), please reply and I’ll switch, thanks!”

That gives them a prompt to resolve the ambiguity.

7 Likes

I do the same, FWIW. And a comment something like “unless you meant the [other thing]”.

2 Likes

That’s what people are usually doing, but it’s far from being true all the time, just yesterday there’s a photo with multiple organism and observer’s finger poinitng to one of them, but cv caught the spider in the background and ided it, with no explanaition added that it’s a separate obs for the spider, I have little doubt in the fat that user wanted an id for something they’re pointing at. Cv also can be erratic and id big picture of flower with tiny thing in the corner as an insect (of course wrong one from the picture) or id the flower, you can’t know what the observer wanted really.

1 Like

For that ambiguity I don’t ID - I leave a comment - did you want the bug or the flower?

Otherwise if the observer has chosen ‘the bug’ then I tag along.

Until they come back, no actually the flower … and we fall in line. Eventually.

2 Likes

This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.