Your samples aren’t sent to any servers, unless you save your project to Google Drive — and even then, it’s in your Google Drive, so that sample data is still yours. When you train the model, it trains in your browser tab without sending anything to any servers.
And version 1 of the Teachable Machine did support output as image, sound effect, or recorded speechHi Matti and Nicolas,
Great tool! It seems to be a great alternative as an initial hands-on activity to teach students image classification, as it simplifies GTM and brings in the fun component of connecting the result to a sound. It has the further advantage of privacy by not sharing any of the images used by the students.
An improvement suggestion would be to add a performance evaluation of the trained model in a future version.Best regards,
Christiane
......................................................................................................................................................
Prof. Dr. rer. nat. Christiane Gresse von Wangenheim, PMP
Software Quality Group - GQS
National Institute for Research and Technology on Digital Convergence - INCOD
Department of Informatics and Statistics - INE
Federal University of Santa Catarina - UFSC
88049-200 Florianópolis - SC
Brazilhttp://www.inf.ufsc.br/~gresseIniciativa Computação na Escola
.....................................................................................................................................................
"Todos os alunos em todas as escolas devem ter a oportunidade de aprender computação.".......................................................................................................................................................................On Fri, Jun 9, 2023 at 3:30 PM Matti Tedre <xxxxxx@uef.fi> wrote:Hi all,
we developed an educational tool that works and looks like Google's
Teachable Machine 2 (and is compatible with models made with it) -- but
to which you can define no-code actions to each classifier result and
then deploy the finalized tool as an "app" (in a new browser tab or to
friends' devices through p2p sharing). For instance, to a classifier
result "dog" you can bind actions like record a "woof!" sound, show a
picture of Spotty, and print text "hey spotty!" as outputs. You can
then deploy the "app" to friends' phones which will then all woof and do
those other things when they see Spotty.
https://tm.generation-ai-stn.fi/
This tool is classroom + GDPR safe: It does not collect or store any
personal data about your use of this application. It does not use
cookies and there is no tracking of activity. Training the model is 100%
done on your local device (you can unplug your network cable once the
page loads). Your sample data and models remain on your device unless
you choose to share them. When sharing, the data are shared through
peer-to-peer connection and not stored in our servers, and becomes
unavailable to others if you close your browser tab.
The tool is available in three languages: English, Finnish, and
Brazilian Portuguese, with more to appear in the future. It is aimed at
those without a coding background. Tutorial video will be there soon.
It's been classroom-tested a lot this spring, and while there are some
issues with peer-to-peer sharing under some network configurations, it's
stable. Bug reports, requests, questions can be mailed to Nicolas Pope
(xxxxxx@uef.fi) or posted to GitHub (https://github.com/knicos/genai-tm).
Let us know what you think!
--GenAI team
https://www.generation-ai-stn.fi/
To unsubscribe from this list please go to https://www.simplelists.comTo unsubscribe from this list please go to https://lists.aaai.org/confirm/?u=HgcrDKbrGvVfL7TMw7K0LcD05I6H3rlD