[Nhcoll-l] AI-based bone identification

[Sullivan, Steven]

As part of a class assignment, my students occasionally use their
preferred AI to ID specimens, including bones.  Results have been very
mixed.

Sometimes the AI is surprisingly accurate.  Results are best if the photo
is clear, with a plain, contrasting  background  and the specimen is
presented with minimal distortion from perspective or other compositional
errors.  However, it is possible for three students to use the same photo
and get different IDs, even if they use the same AI.  Similarly, simply
asking for an accurately IDed bone to be IDed again after some intervening
IDs may result in a misID the second time.

We have a giraffe metatarsal scrimshawed with cape buffalo images that is
often, but not always, IDed correctly to species, even in images without
scale or other reference objects.  In contrast, deer cow, and other
metacarpals are often IDed as the correct bone but not the correct
species.

The more articulated or jumbled the collection, the less accurate the ID.
(eg articulated hands or paws are less often in the ballpark than are
articulated ungulate lower limbs.)  With these specimens, while the
anatomical ID may be more or less accurate, the species is pretty random,
trending towards domesticated species or those that every little American
kid learns (eg fox or deer, never jackal or antelope).  Similarly, the more
complex the bones, the less accurate the ID. (eg Articulated cheetah
vertebrae may be IDed as a fish spine).

Since this is an optional part of an assignment, I have not approached any
of this systematically but it seems that the more bones in the image, the
less useful the AI is.  Image quality and composition always matters but,
with bones, this is far less important than insects, plants, and field
observations.  At this point, I'd say bone ID is correct often enough to be
a neat thing that computers can do.  It is incorrect often enough (while
still seeming to assert accuracy) that it is not very useful in helping
people learn.  Nor does it provide enough discussion or rationalization of
its assessment to help knowledgeable people use it to make decisions.  In
contrast, AI is accurate enough at IDing manufactured items, even from
pretty bad images, that I have adapted my assignments.

Though none of the readily available, off-the-shelf AIs are really useful
in my experience on bones, I can imagine that some kind of AI could be
trained to be really accurate.

--Steve

On Tue, Feb 3, 2026 at 10:21 AM Battermann, Nora <battermann at snsb.de> wrote:

> Dear all,
>
>
> Apologies for cross-posting.
>
>
> I’m interested to find out whether anyone has experience using AI-based
> tools/approaches to identify bones or skeletal elements from photographs.
>
>
> The specific context is zooarchaeological assemblages, and I’m
> particularly interested in whether AI has been used (successfully or
> unsuccessfully) to determine if images of mixed bones/fragments contain
> certain element types (e.g., phalanges, teeth, long bone fragments, etc.),
> rather than to species level.
>
> If you’ve worked with any relevant software, models, workflows, or pilot
> projects—or know of published or ongoing work in this area—I’d be very
> grateful for any pointers or insights.
>
>
> Many thanks in advance,
> Nora
>
>
> ---------
> Dr. Nora Battermann
> Koordination Sammlungsmanagement
>
>
> SNSB – Staatliche Naturwissenschaftliche Sammlungen Bayerns
> Zentrale Einrichtungen
>
> Tel.: +49 (0)174 2762445
>
> www.snsb.de
>
>
> _______________________________________________
> Nhcoll-l mailing list
> Nhcoll-l at mailman.yale.edu
> https://mailman.yale.edu/mailman/listinfo/nhcoll-l
>
> _______________________________________________
> NHCOLL-L is brought to you by the Society for the Preservation of
> Natural History Collections (SPNHC), an international society whose
> mission is to improve the preservation, conservation and management of
> natural history collections to ensure their continuing value to
> society. See http://www.spnhc.org for membership information.
> Advertising on NH-COLL-L is inappropriate.
>


-- 

<http://miamioh.edu/>

Steven M. Sullivan

Director | Hefner Museum of Natural History
<http://miamioh.edu/cas/academics/centers/hefner-museum/>

Miami University

100 Upham Hall

100 Bishop Circle

Oxford, OH 45056

Museum: 513-529-4617

Cell: 708-937-6253


The Museum is open to the public weekdays 9-4, free admission.

Support the Museum today!
<https://securelb.imodules.com/s/916/lg21/form.aspx?sid=916&gid=1&pgid=6010&cid=11236&dids=292&bledit=1>


*Connecting you to the nature in your neighborhood...and the world.*


Miami's many museums and collections
<https://www.miamioh.edu/cca/museums-miami/index.html> provide unique,
cross-disciplinary opportunities for students, educators, and the public.


The Hefner Museum recognizes the Myaamia and Shawnee people who, along with
other indigenous groups, were the first stewards of this land's
biodiversity. <http://miamioh.edu/diversity-inclusion/land/index.html>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.yale.edu/pipermail/nhcoll-l/attachments/20260203/a7a85c1f/attachment.html>