A new Computer Vision Model including 1,403 new taxa in 32 days
We released a new computer vision model today. It has 68,853 taxa, up from 67,553.
This new model (v1.5) was trained on data exported exported last month on November 13th and added 1,403 new taxa.
Taxa differences to previous model
The charts below summarize these 1,403 new taxa using the same groupings we described in past release posts.
By category, most of these 1,403 new taxa were insects and plants
Here are species level examples of new species added for each category:
- Other Animalia
- Mollusca
- Aves
- Fungi
- Reptilia
- Plantae
- Insecta
- Actinopterygii
- Amphibia
- Arachnida
- Mammalia
Click on the links to see these taxa in the Explore page to see these samples rendered as species lists. Remember, to see if a particular species is included in the currently live computer vision model, you can look at the “About” section of its taxon page.
We couldn't do it without you
Thank you to everyone in the iNaturalist community who makes this work possible! Sometimes the computer vision suggestions feel like magic, but it’s truly not possible without people. None of this would work without the millions of people who have shared their observations and the knowledgeable experts who have added identifications.
In addition to adding observations and identifications, here are other ways you can help:
- Share your Machine Learning knowledge: iNaturalist’s computer vision features wouldn’t be possible without learning from many colleagues in the machine learning community. If you have machine learning expertise, these are two great ways to help:
- Participate in the annual iNaturalist challenges: Our collaborators Grant Van Horn and Oisin Mac Aodha continue to run machine learning challenges with iNaturalist data as part of the annual Computer Vision and Pattern Recognition conference. By participating you can help us all learn new techniques for improving these models.
- Start building your own model with the iNaturalist data now: If you can’t wait for the next CVPR conference, thanks to the Amazon Open Data Program you can start downloading iNaturalist data to train your own models now. Please share with us what you’ve learned by contributing to iNaturalist on Github.
- Donate to iNaturalist: For the rest of us, you can help by donating! Your donations help offset the substantial staff and infrastructure costs associated with training, evaluating, and deploying model updates. Thank you for your support!
Σχόλια
Very cool! But protozoa link goes to a global stats page.
Got 12 species in there, glad that photographing every Ulochlaena hirta specimen led to incluing it in a model!
good catch - no protozoa this time, I removed it
Awesome! Waiting for Platanthera pallida to get added(it already has over 100 photos and 50 RG photos).
That's an interesting case, @yayemaster.
With only 8 observers (70% of observations coming from a single user) and 11 identifiers (65% of identifications coming from a single user), I could make the case that our community as a whole doesn't understand that species well enough for our vision system to be able to reasonably incorporate it into the model. It is certainly the case that the diversity of camera and photography techniques will be underrepresented for this taxon.
So even when it crosses the boundary, it may be the case that our models underperform on this taxon. It probably will work fine for you, since you took 70% of our photos of this taxon, so our model will likely learn to associate this taxon with both the visual features of the organisms as well as with your camera and your photography habits (framing, preferred zoom level, time of day you go on hikes, lighting preferences, etc). But it may not work as well for other users, with other habits. Machine learning models are notoriously good at finding these kinds of short cuts in the data.
This update is a couple of days earlier than expected, great work! Are beside the new taxa added, also already existing taxa in the model improved because for example more photographs are available?
Yes, @rudolphous , that's correct.
I'm wondering about Ericameria fasciculata (https://www.inaturalist.org/observations?locale=en&place_id=1921&preferred_place_id=1&quality_grade=research&taxon_id=76882&verifiable=any); it has 64 RG observations, with 37 observers, 100+ RG photos. I did do some identifying recently, so maybe it was just under the threshold.
Does the model take # of observers or their individual # of observations into account as well?
Thanks for all you guys do, this is a really awesome project to be a part of in some small way!
Love it and it is still a cool feature that one can now look those new species up. Interesting to see what and where was added
It's great that the computer vision model is continuously improved but I still miss hybrids. Are there plans to add them back into the model in the near future?
Great update, thanks for sharing
The rest of us try too :-)
keep it up folks! great job
Great update
Noice job guys keep it up!
I have no idea how difficult it is (or how expensive in computertime) but is it possible to give directions, tips, hint to the Computer Vision to
-Recognise more species on one photo
Recognize the status of the animal (alive, dead, caterpillar, flying, young, adult, eating, sleeping, flowering, in bud)
Will it still be possible in the future to extend the model every month with around 1000 species or wil there be an optimum, a kind of barriere. asymptoot ?
Προσθήκη σχόλιου