Jev in ActionInteractive use cases
☆ Star on GitHub
GESTURE-CONTROLLED WEB TOOLS

A little motion. A clear intention.

Your hands provide the signal. JEV chooses the tool.

Client-side vision · live JEV
01 / MAKE YOUR MOVECamera optional
ON-DEVICE VISION
Hands in the frame.
Ideas in motion.

Camera off · simulations ready

Or try an observation below ↓
02 / INTERACTIVE CANVAS1 object · 1 target
Interactive gesture canvasA violet card and a fixed drop target. Pinch over the card to request a grab, move, then release inside the target. Drop here Pinch to grab

Move the blue fingertip cursor onto the card, pinch or close your fist and wait for JEV, then move to drag. Open your hand to drop. To reset, make a fist and open it twice within 3 seconds. Holding an open palm does nothing.

Tools light up when JEV selects them.

NO CAMERA? NO PROBLEM.

Try Gesture Observations

Same structured input. Same live JEV decision flow. An API key is required.

Click to simulate
SIMULATED TRACKING

Point + hold targets the card. Simulate grab uses live JEV; movement and release then run locally. Close + open twice simulates the completed reset sequence. Simulating stops the camera so the two inputs never compete.

Webcam → Hand Landmarker → Gesture features → Structured observation → JEV → Selected tool → Canvas
Inspect the JEV request & response

Request

No request yet.

Response

No response yet.

Camera frames and hand landmarks stay on your device. Only structured observations are sent to JEV through this server. Hand tracking assets load on demand.

Connect JEV

Use your existing playground key or add one here.

Keys are sent through this server to the provider, not stored on the server. By default, browser storage lasts for this session.