Limit computer vision suggestions for observations of humans

If the computer vision system returns human as one of its options, or if need be if it returns that the ‘pretty sure’ answer is genus Homo, then no other options should be presented to the user.

Don’t fill it out with any other choices. Doing do would hopefully cut down on the number of ‘joke’ ID’s that come in from students (and others) taking photos, running the ID and getting ‘funny’ choices they can select.

I have had the computer suggest Homo sapiens at the top of the list for something random, such as a moth, but then it has the correct identification below in the list. A couple of times, it seemed like the computer is just determined to get me to submit an observation of a human. This may be a good idea to implement, but I think it would need a way for people to use the identification suggestions even if the computer is convinced that there is a human in the picture.

1 Like

I have no evidence, but I don’t think the joke IDs are particularly fueled by the computer vision program - They were being added long before it existed.

Another trouble is many observations of things are photos of people holding a specimen (eg a reptile), might it be more useful for the CV program to not bother with presenting ‘human’ at all as an option, but just the secondary suggestions …

7 Likes

When working through State of matter Life the most common misidentification of Homo sapiens is as Agaricus bisporus. This is a result of the computer vision option.


I think this is a temptation. I have no idea how hard the coding to do this would be, but when the computer vision lists Homo sapiens as the first possibility, top of the rank order list, no other possibilities should be offered by the computer vision. If the image is truly of some other species, the user would have to type in that name manually. From what I have seen the computer vision is really accurate on picking up Homo sapiens, and this might cut down on the joke IDs by removing the temptation to choose a mushroom or other species. This would not end the problem, but would certainly encourage identifying the image as Homo sapiens.

3 Likes

https://forum.inaturalist.org/t/limit-computer-vision-suggestions-for-observations-of-humans/2067

1 Like

We discussed this as a team and decided that we didn’t want to take a stance in saying that we “know” for sure what the ID is. We did, however, remove the Homo genus suggestion from the list if Homo sapiens is the top result, as there was a pattern of students choosing it as a possible joke/insult.

8 Likes

Then this request as well as the original one linked to it should be closed to free up votes.

Ideally the help page and/or curator guide should also be updated to specifically note that users who choose an obviously wrong ID for a human as a result of it being offered by the CV are not eligible for suspension to help document /reslove the constant flow of flags that come from this (as a curtator, I am perfectly fine updating how-to pieces of the document, but site policy stuff should be written by site employees)

1 Like

5 posts were merged into an existing topic: Should the Curator Guide be updated in regards to obvious wrong IDs on observations of humans?

That’s the plan, I just wanted to make sure you had a chance to reply.

Your curator guide suggestion would probably be best suited for the Curator category, so I’m going to start one there and move the relevant replies to it.

1 Like