Control your computer with hand gestures using your webcam! A low-latency, customizable gesture control system powered by MediaPipe.
🏆 Built at HackCamp 2025 by nwPlus - A hackathon project focused on accessibility and hands-free computing.
- 🖱️ Move Cursor - Control cursor with open hand
- 👆 Click Actions - Pinch for left click, configurable hold duration
- 🖱️ Right Click - Custom gesture for right-click
- ⬆️⬇️ Scroll - Thumbs up/down for scrolling
- ⏸️ Toggle Control - Pause/resume gestures on the fly
- 🎛️ Real-time Settings - Adjust all parameters without restarting
- 🔄 Gesture Remapping - Assign any gesture to any action
- 🔒 Window Lock - Click-through camera overlay option
- ⚡ Low Latency - Optimized for real-time performance (~30-50ms)
- OS: Windows 10/11 (64-bit)
- Python: 3.11 or higher (for source)
- Webcam: Any USB or built-in camera
- RAM: 4GB minimum, 8GB recommended
- Download the latest release from Releases
- Extract
AccesiGesture-vX.X.X-Portable.zip - Run
AccesiGesture.exe - Allow camera access when prompted
# Clone the repository
git clone https://github.com/dlt87/Accesigesture.git
cd Accesigesture
# Create virtual environment
python -m venv venv
venv\Scripts\activate
# Install dependencies
pip install -r requirements.txt
# Run the application
python main.py| Gesture | Default Action | Description |
|---|---|---|
| ✋ Open Hand | Move Cursor | All fingers extended |
| 👌 Pinch | Left Click | Thumb + Index finger touch |
| 🤌 Pinch Middle | Right Click | Thumb + Middle finger touch |
| 👍 Thumbs Up | Scroll Up | Only thumb extended, pointing up |
| 👎 Thumbs Down | Scroll Down | Only thumb extended, pointing down |
| 🤙 Toggle | Pause/Resume | Thumb + Index + Pinky extended |
Pro Tip: Hold pinch for 0.15s (adjustable) to enable click-and-drag mode!
Access the settings panel to customize:
- Smoothing Factor (0.0-1.0): Higher = more responsive, lower = smoother
- Presets: Quick options (Smooth, Balanced, Responsive)
- ROI Boundaries: Define tracking area on screen
- Pinch Threshold: Sensitivity for click detection
- Pinch Duration: Hold time before drag mode activates (default: 0.15s)
- Fist Cooldown: Delay between right-click actions
- Scroll Speed: Adjust scroll distance
- Detection Confidence: Initial hand detection threshold
- Tracking Confidence: Continuous tracking sensitivity
Customize any gesture to perform any action via dropdown menus in the settings window.
- Check camera permissions in Windows Settings
- Ensure no other app is using the camera
- Try changing
cv2.VideoCapture(0)to(1)in main.py
- Lower camera resolution (already optimized to 320x240)
- Increase smoothing factor in settings
- Close other resource-heavy applications
- Ensure good lighting conditions
- Ensure good lighting conditions
- Position hand within camera frame
- Adjust detection confidence sliders
- Keep hand between 30-60cm from camera
- Use plain background for better detection
- Windows are automatically positioned on startup
- Camera window: Top-left (20, 20)
- Settings window: Top-right with margins
- Use "Lock Window" feature for click-through camera view
- Optimal Distance: 30-60cm from camera
- Lighting: Bright, even lighting works best
- Background: Plain backgrounds improve detection
- Hand Position: Keep entire hand visible in frame
- CPU Usage: ~5-10% on modern processors
- Multi-hand gesture support
- Custom gesture recording
- Gesture profiles/presets
- Cross-platform support (Linux, macOS)
- Voice command integration
- Eye tracking integration
- Accessibility features for motor disabilities
- HackCamp 2025 - Created during the nwPlus HackCamp 2025 hackathon
- MediaPipe - Hand tracking and gesture recognition
- OpenCV - Computer vision processing
- PyAutoGUI - Mouse and keyboard control
- Tkinter - GUI framework
- Issues: GitHub Issues
- Discussions: GitHub Discussions
- Latest Release: Download
- v1.0.1 - (Official Hackathon Submission) UI improvements, pinch duration slider, window positioning
- v1.0.0 - Initial release with core gesture control features
Made with ❤️ for accessibility and hands-free computing
Built at HackCamp 2025 by nwPlus