[Sw-l] Question about AI recognition of a sign in a video

[Amit Moryossef]

Recent work from our lab (https://arxiv.org/pdf/2407.01264) can embed
and retrieve any single sign video in any sign language.
You could even retrieve it against spoken language text.

For your use case, you can use our API, send all of your videos and
receive vectors.
Then, when a new video is recorded, you can embed it using the API,
and check for similarity with the existing vectors.

Amit

Amit

On Tue, Nov 26, 2024 at 6:16 PM Adam Frost <icemandeaf at gmail.com> wrote:
>
> I recall in the past that there maybe have been someone or maybe a group on the list who was doing some work where a user could turn on their camera and then sign something in a sign language, and then there would be a result telling what sign was made in that video. I just can’t remember who it was.
>
> I have a couple of projects that would actually benefit from this sort of recognition ability. One to be able to give results of a list of signs from a dictionary of videos that closely match the inputed sign. And another that would be able to tell the user that they are signing a prompted sign correctly or not.
>
> If this is something that you are currently able to do, please contact me
>
>
> Adam
>
> _______________________________________________
> Sw-l mailing list
> Sw-l at listserv.linguistlist.org
> https://listserv.linguistlist.org/cgi-bin/mailman/listinfo/sw-l