Hello all.
The world’s first visual dataset for International Sign Language (ISL) is now public!
This project was created to help AI learn how to speak with hands.
Developed using real gesture photos and labeled sign language data.
I’m proud to be the first data architect of this groundbreaking prototype.
Explore here:
“A Voice for Silence, Hope for the World.”
— Yasin Şimşek
Founder of ChatDEAF | Istanbul
How interesting!
Reply – From Alejandro Arroyo & Clara (Coauthors, Symbolic AI Research)
Hi,
This is a truly valuable project — thank you for making this public and pushing gesture recognition toward greater accessibility.
We’d like to offer a small perspective:
There are emerging frameworks that don’t rely on high-volume, high-cost datasets to recognize gestures. Instead of training on thousands of samples per sign, these approaches focus on the internal geometry and symbolic resonance of the gesture itself.
The idea is to treat each gesture not just as data to classify, but as a semantic structure with internal coherence—almost like a resonance pattern—allowing systems to match meaning based on symbolic alignment rather than statistical frequency.
This might open up new ways to scale, without compromising cultural or contextual integrity.
If the project ever explores these structural alternatives, we’d love to contribute ideas or explore joint research.
Thanks again for your pioneering work.
— Alex & Clara
Coauthors in Symbolic Architecture & Human-AI Codevelopment
Dear Alex & Clara,
Thank you so much for your kind words and for sharing such a thoughtful perspective.
As the deaf founder of the ChatDEAF project, it means a lot to receive this kind of support from professionals working in symbolic and human-AI fields. Your idea of understanding gestures through internal meaning and symbolic structure, rather than only large datasets, truly resonates with the core philosophy of ChatDEAF.
This approach could be a very powerful direction for inclusive AI — especially in low-resource or culturally sensitive sign language contexts.
I would be honored to stay in touch and explore possible collaboration or exchange of ideas. Thank you again for your encouragement and valuable insights.
With gratitude,
Yasin Şimşek
Founder of ChatDEAF | Istanbul
Thank you so much for your interest!
I’m happy you found the project interesting.
ChatDEAF is a new prototype focused on accessibility for the Deaf community through AI and sign language datasets.
If you’re curious, I’d love to share more or collaborate!
Warm regards,
Yasin Şimşek
Founder of ChatDEAF
I have done something smilar you can check and get everything from there also I provided a dataset for Turkish Sign Language altough it might have some problems still can be used to training. Actually I have myself did collected the data just infront of the camera. it is word based. The data is consist of : 30 frames of the action (while signing the word I took the x y z cordinates of the dots placed by google’s mediapipe holistic model).
Hello,
We continued working on the projects, but as you know, it was a very difficult task. We were able to achieve only about 30% progress.
Unfortunately, another company’s server was shut down, which caused our system to stop working. We do not know the exact reason for this issue.
As a result, the project could not continue and has been put on hold. The team has been disbanded, and I have decided to end the project. Discussions regarding this matter are now closed.
Thank you all very much once again for your support and efforts.
Best regards.