CYRA Lab · Module 04
Silent Talk
Understand each other without a word.
Talk to CYRA with hand signs: wave to say hello, tell it how you are with your thumb, show a number with your fingers. CYRA reads every sign and answers in writing and, if you like, out loud.
- Runs on your device
- ~11 MB on first start
- Model: MediaPipe Gesture Recognizer · hand skeleton (21 points)
How to try it
Turn on the camera and bring your hand into the frame; CYRA draws its skeleton.
Wave: CYRA says hello and asks how you are.
Thumb up or down, then one to five fingers, and finally the “I love you” sign: the conversation moves on step by step.
Your camera permission is used only on this page and only when you turn it on.
Your image never leaves your device: no frame is recorded or sent.
Sign vocabulary
- Wave · hello: current step
- Thumb · how are you?
- Fingers · a number
- The I-love-you sign
Sign I see
—
Open fingers
—
Signs CYRA knows
- Open palm
- Thumb up
- Thumb down
- I love you
- Victory
- Pointing up
- Fist
CYRA log
- Once you turn on the camera, we start talking here.
Honest note: this module is not a sign-language translator; it is CYRA's small vocabulary of hand signs. The ready-made Turkish Sign Language (TID) models we could find are either unlicensed or trained on research-only data. Training a TID model on our own dataset is a next step we are considering.
Behind the scenes
- The hand model returns 21 joint points and seven ready-made signs every frame: open palm, fist, thumb up and down, victory, pointing up, I love you.
- We count the fingers ourselves from how far the joints are from the wrist; scaled to the size of the palm, so the result is the same whether your hand is near or far.
- One-frame flicker does not count: a sign is accepted after half a second, a number after one second, without interruption.
Where it helps
- Touchless control: command lights, blinds or a pergola without touching a panel with wet or dirty hands.
- Noisy places: confirm with a hand sign in workshops and on site where voice does not work.
- Accessibility: interfaces that need neither speaking nor hearing and answer in writing.
Example use ideas; we design the right one for your project together.
Next module
Teach CYRA
Show a sign, and it learns it in seconds.
What would you like your project to sense?
Motion, objects, people or surroundings: tell us what you need and let's design the embedded solution together.
Models: Google MediaPipe Gesture Recognizer, Face Landmarker and EfficientDet-Lite0 (COCO dataset), Apache-2.0 licensed. Spoken replies use your browser's own voice.