Learning more about iNat's computer vision

I would like to read the releasenotes on computervision. What has changed and improved since june 2017 in the CV model ?
I did not know that only the first photo was used for CV in stead of the best photo.

And about the performance it costs. Is it bad for iNaturalist to use the CV by default or is it better to use it on request ? I heard 10 graphic cards in a server was enough to get a good response and i was wondering if such a thing changes in time.

No, see this related feature request:
https://forum.inaturalist.org/t/toggle-computer-vision-suggestions-on-off-on-website/1117

1 Like

I didn’t know that either. I seem to recall that the original iNat CV demo page allowed for multiple photos, so assumed it would evaluate all and combine the results.

(Just checked https://www.inaturalist.org/computer_vision_demo, and I remember wrong - it only allowed one photo)

IT explains the bad results sometimes. The other CV all do multiple fotos as far as i know.(Naturalis and the French one, PlanetNet)

https://forum.inaturalist.org/t/species-suggestions-for-the-wrong-continent/789/15?u=ahospers

https://forum.inaturalist.org/t/species-suggestions-for-the-wrong-continent/789/15?u=ahospers

To me I feel like once the computer vision has a species well known, it is really good at knowing what is that species, but it’s really bad at knowing what is not that species.

So for example if you post a photo of Toxomerus marginatus (a small common colourful pollinator in North America), it will get it right away almost every time, but if you post a completely different looking small fly, bee, or wasp, it might also call it Toxomerus marginatus just because it’s also a small pollinating insect. This is problematic when most small insects are too difficult to identify to get enough species identifications for the computer vision to know about (20 RG observations by 20 different users I believe).

1 Like

I didn’t see this page linked in this topic, so dropping it in:
https://www.inaturalist.org/pages/computer_vision_demo

Though, I’m not sure how much has changed in the threshold/process since it was last substantially edited in late 2017.

I thought the Taxonomy was not suitable for advising a higher ranking taxon (family) if it is not sure about the species.
Here (actualy the firm seems to be a few hunderd meters from my workplace) the are busy with counting the few insects that are left: https://phys.org/news/2019-05-automatic-insect-identification-grasp-biodiversity.html

Sorry, I’m not sure I understand what you mean. Newer iterations of computer vision are ranking some genera and other taxonomic ranks higher than species, e.g. Trombidium holosericeum used to be the top suggestion for basically any tiny red terrestrial arthropod, but it’s now often ranked lower in the results than genus and family:

1 Like

I never knew it could work that way…but it is only on the web interface ? Not on the IOS app ?

It’s new-ish. Maybe you need to update your app?
https://www.inaturalist.org/blog/25510-vision-model-updates

See this related feature request to allow use of computer vision on each photo in an observation:
https://forum.inaturalist.org/t/use-computer-vision-on-each-photo-in-an-observation/4210

Also, I added a few common questions and answers to the FAQ page on the iNat website about computer vision: https://www.inaturalist.org/pages/help#computer-vision If there are others you can think of, adding them here might be helpful so that the staff or curators can consider other additions to the FAQ.