Home
Jarvis Gesture Control Arduino SaaS
About Devlog Contact
← All projects

Project 02  /  Accessibility

Gesture
Control

Hands-free Mac control through hand and face gestures. Built for accessibility — designed for people with mobility limitations who need a better way to use a computer.

Active build PythonMediaPipeOpenCVAccessibility
Overview

What it does

A real-time computer vision system using MediaPipe and OpenCV that translates hand and face gestures into Mac system actions — desktop switching, volume, scrolling, playback, and more.

Runs as a lightweight background process on an M3 MacBook Air. The M3 neural engine handles MediaPipe inference efficiently with no noticeable performance hit.

Wave gesture toggles brainrot mode for hands-free TikTok/YouTube scrolling using head nods. Safety lockouts prevent accidental triggers when you're touching your face.

Gesture map

Input
Gesture
Action
Right hand
Swipe left/right
Switch desktop
Right hand
Fist
Play / pause
Left hand
Pinch
Volume
Face
Eyes closed 5s
Sleep
Face
Mouth open 2s
Caps lock
Face
Head nod (brainrot)
Scroll
Right hand
Wave
Toggle brainrot mode
Gallery
Demo video / GIF
Hand tracking overlay
Face gesture detection
Mudra detection mode
Code structure
Tech Stack
MediaPipe
Hand + Face tracking
OpenCV
Camera feed
PyAutoGUI
System actions
Python
Core logic
M3 Neural Engine
Inference
DJI Action 2
Camera input
Roadmap

Current state

Full gesture system working. Both hand and face detection running simultaneously. Safety features (hand lockout, face/hand distance check) implemented and tested.

Separate mudra_detect.py module detects 13 Bharatanatyam mudras with confidence scoring and snapshot saving.

Next steps

  • Ollama AI script builder (v3.0)
  • DMG distribution for easy install
  • Research paper on control systems + AI
  • Professor outreach for collaboration
  • Accessibility org partnerships
View on GitHub Get in touch