Your webcam finally earns its keep as a mouse

What it does The project maps hand gestures from a standard webcam to full mouse control—cursor movement, left and right clicks, scrolling, drag-and-drop, and even volume and brightness adjustments. A bundled voice assistant named Proton adds spoken commands for Google searches, file navigation, and toggling the gesture camera on or off. It is built around MediaPipe hand-detection models and OpenCV, and the authors explicitly limit support to Windows.

The interesting bit Instead of extra hardware, the system offers two input modes: bare-hand tracking via MediaPipe or a glove of any uniform color. The voice assistant can sleep, wake up, and hand control back to the gesture module, so the two input channels cooperate rather than compete.

Key highlights

• Maps static and dynamic hand gestures to cursor control, clicking, scrolling, and multi-item selection
• Supports scroll, volume, and brightness controls where the speed scales with the distance of a pinch gesture
• Includes a voice assistant, Proton, that opens Google searches, navigates local files, and toggles gesture recognition
• Runs on a standard webcam without additional hardware, according to the README
• Offers two recognition modes: direct hand tracking or color-glove detection

Caveats

• Windows-only; the README explicitly limits support to that platform
• Hard-targets Python 3.8.5, which may frustrate anyone on newer releases
• Exiting the voice assistant requires manually closing the GUI window afterward

Verdict Worth a look if you want to experiment with computer-vision input pipelines on Windows or need a starting point for a MediaPipe-based HCI project. Skip it if you need cross-platform support or a polished, production-grade input driver.