Gesture-Controlled Computer Interface in Python
A real-time, touchless computer vision control interface in Python using OpenCV, MediaPipe 21 hand landmarks, EMA cursor smoothing, and…
A real-time, touchless computer vision control interface in Python using OpenCV, MediaPipe 21 hand landmarks, EMA cursor smoothing, and multi-frame temporal stability voting. Translates hand gestures into OS mouse and media commands with sustained drag-and-drop, pinch clicks, two-finger scrolling, and system volume control.
- Full Python OpenCV and MediaPipe 21 skeletal hand landmark tracking pipeline operating at 60 FPS
- Multi-frame temporal voting engine (Challenge Clue) requiring quorum consensus before triggering actions to eliminate single-frame classification noise
- Exponential Moving Average (EMA) cursor smoothing and deadband filter removing hand jitter without latency
- 7 distinct gestures: Pointing (cursor move), Pinch (click), Sustained Pinch (drag & drop), Middle Pinch (right click), Two-Finger (scroll), Shaka (volume), and Fist (standby)
- Automated debounce cooldown timer preventing accidental double-firing on discrete click events
- Dynamic chirality adaptation supporting both Left and Right hands with mirror feed correction
- Hot-reloadable local JSON configuration (gestures_config.json) with custom thresholds and keyboard controls
- Interactive Web Showcase Studio featuring live webcam feed, synthetic hand simulator, virtual desktop sandbox, and 8-point automated diagnostic suite
Built with: Python 3.12, OpenCV, MediaPipe Hands, 21 Skeletal Landmarks, Temporal Multi-Frame Voting, Exponential Moving Average (EMA), PyAutoGUI / OS Automation, Tremor Deadband Filter, Local JSON Configuration, Left & Right Hand Chirality, HTML5 / CSS3 / Vanilla JS, Google Fonts (Sora, Outfit, Chivo Mono)