AutoMagic Vision – Hand Tracking & Gesture Control

Python computer vision project for hand tracking, gesture-based controls, and adaptive mouse automation. Uses OpenCV and MediaPipe for real-time gesture recognition.

RoleLead Computer Vision Developer

TimelineMar 2026

TeamPersonal Project

Tech Stack5 Technologies

Mission Brief

AutoMagic Vision is an AI-driven computer vision automation suite in Python. Leveraging Google MediaPipe and OpenCV, it performs real-time 21-point 3D hand landmark estimation from standard webcams. It maps fine motor finger gestures to OS-level mouse navigation, click triggers, drag-and-drop actions, volume adjustments, and keyboard shortcuts without requiring physical hardware sensors.

Key Features

Landmark Tracking & Automation

MediaPipe 21-Point Model: High-accuracy real-time finger joint and palm tracking
Virtual Mouse Controller: Cursor movement tracking index finger position with exponential moving average smoothing
Gesture Gestalt: Pinch gestures trigger clicks, open-palm navigates windows, and two-finger scroll adjusts volume

Engineering Chronicles

Hand landmark jitter caused the virtual mouse cursor to tremble on small motor movements.

SolutionImplemented an exponential weighted moving average (EWMA) and dynamic deadband filter that stabilizes subtle tremors while maintaining high-speed responsiveness.