[semantics-public] [x3d-public] URL resolution problem on web3d.org - for X3D Ontology Assets; X3D Semantic Web meeting minutes

Don Brutzman brutzman at nps.edu
Mon Feb 3 21:33:45 PST 2020

Wow, some intense (borderline astonishing) ideas in no time flat!  Hello out there!  8)


On 2/3/2020 5:24 PM, GPU Group wrote:
>  > i wrote:>  > Might anyone else be interested in submitting a semantics-related paper, perhaps we can organize a semantics session?  We have 19 days.
> IDEA: render scenes found to an image/screenshot and train AI with (screenshot,semantics) tuples, so when AI sees a screenshot of something, it can tell what's in the scene;.
> -
> Doug

What I really like about this idea is that the idea of a limited training set almost becomes moot.  With CG rendering, with corresponding known semantics, and a variety of viewpoints/lighting/filters/effects/etc. a system could create an arbitrarily large synthetic training set for any set of objects.

That could be a major project and a significant advance in computer vision.


On 2/3/2020 5:37 PM, Felix Hamza-Lup wrote:
> Another idea related to _*medical X3D and associated ontology*_. I have a few ideas but I'm under heavy workload this month.
> Building an Ontology for various surgical tasks .... dental procedures... for simulation and training purposes.

HAnim 2 includes
- a number of baseline aliases for joints, segments, sites;
- these could be matched to any number of medical ontologies through triple correspondences,
- addition of HAnimMotion node (patterned after BVH) for motion animation is an enabler for building a library of surgical/dental tasks and procedures.

Suggested path forward:
- mature our tools further, adding OBJ/Shape/STL support for 3DMD and NIH device printing,
- implement security in tools using XML Digital Signature (authentication) and XML Encryption (privacy),
- combine these to establish baseline for X3D supporting medical records,
- voila, any work on medical ontology has approximately everyone to apply results to.


On 2/3/2020 6:00 PM, Nicholas Polys wrote:
> Apologies, these meetings are not on my calendar apparently

Mondays 08-09 pacific (like other working group calls)

> but
> I love this idea ~
> my Master's student last year did a study with real scenes:
>     Feed Me: an in-situ Augmented Reality Annotation Tool for Computer Vision
> https://vtechworks.lib.vt.edu/handle/10919/90897

wow again.  CV and AR puts humans in that sophisticated loop.

if we add synthetic training imagery from X3D image snapshots then the possibilities seem approximately unlimited.

>   it was not X3D though... :-]
> maybe we could try w X3D renderings and see how different objects recognition techniques fare...

makes sense

>   ... would they match the X3DOnto description?

The X3D Ontology fully matches X3D version 4, getting updated via X3DUOM each time the working group approves a change.

At some point will autogenerate X3D ontologies for earlier versions of X3D 3.0-3.3, please holler if needed sooner.


These are all potentially huge.  Does anyone have anything that can reach escape velocity over next 18 days?

all the best, Don
Don Brutzman  Naval Postgraduate School, Code USW/Br       brutzman at nps.edu
Watkins 270,  MOVES Institute, Monterey CA 93943-5000 USA   +1.831.656.2149
X3D graphics, virtual worlds, navy robotics http://faculty.nps.edu/brutzman

More information about the semantics-public mailing list