Turn hand gestures into English and Acholi text in real time,
powered by MediaPipe Holistic and a Transformer AI model.
📷
Position Your Camera
Hold your hand 30–60 cm from the camera in good
light. Keep your whole upper body visible so pose landmarks can
be tracked alongside your hands.
🟢
Start Translating
Tap Start Translating and perform your sign
in front of the camera. Every 30 frames the AI model predicts
the gesture and the translation appears automatically.
🌍
Read the Translation
Results appear below the camera in both English
and Acholi (Luo). Tap the copy icon to share the
text, or tap Games in the navigation to practise
Uganda Sign Language interactively.
Quick Help
🟢Start Translating — begins the live 225-float landmark stream to the AI model.
🔴Stop Translating — stops the stream and ends the session.
📶The dot in the header shows WebSocket connection health (green = live).
💡Stay within 60 cm, well-lit, with your full torso visible for best accuracy.
🤖The AI triggers a prediction every 30 frames — results appear automatically.
🌐Tap the globe icon to toggle the Acholi translation on/off.
⚙️Tap Settings in the bottom nav to change the backend server URL when hosting remotely.
🎮Tap Games to practise Uganda Sign Language with SignMaster.
✋SignBridge
Offline
⏳Initialising…
0/30
RHLHPOSE
🔤Translation
EN
Waiting for gesture…
ACH
Kur me twero nyutt coc…
Server LogPipeline Diagnostics
Frame0
225-float frames0
RH zeros0
LH zeros0
WS sent0
Buffer0/30
Settings
Change this when the backend is hosted remotely (e.g. on Render or Railway).
Use wss:// for HTTPS-hosted backends.