A computer vision application that detects hand gestures and counts extended fingers in real-time, indicating whether the count is even or odd. Built with OpenCV and MediaPipe.

• Real-time hand detection and finger counting
• Visual feedback showing even/odd status
• Support for both left and right hands
• Stability tracking for reliable gesture recognition
• Performance metrics display (FPS)
• Fullscreen interactive display

• Python 3.8 or higher
• Webcam or camera device
• Git (for cloning the repository)

git clone https://github.com/ajena555/Hand-gestured-eveodd.git
cd Hand-gestured-eveodd

python -m venv .venv
.venv\Scripts\activate

python3 -m venv .venv
source .venv/bin/activate

pip install -r requirements.txt

python modifiedevenodd.py

Press 'q' to exit the application.

1. The application uses your webcam to capture video frames.
2. MediaPipe's hand detection model identifies hand landmarks in the frame.
3. The algorithm analyzes finger positions to determine which fingers are extended.
4. The total count of extended fingers is calculated and displayed.
5. The application indicates whether the count is even or odd.
6. A stability measure shows how consistent the detection is over time.

• Ensure good lighting conditions for optimal hand detection.
• Position your hand clearly in the camera's view, palm facing the camera.
• Keep your hand movements slow and deliberate for better stability.
• For thumb detection, extend it clearly to the side.
• The application works best when your hand is approximately 12-24 inches from the camera.

Create a file named requirements.txt with the following contents:

opencv-python>=4.6.0
mediapipe>=0.8.10
numpy>=1.20.0

1. Poor detection accuracy:

   • Improve lighting conditions
   • Position hand more clearly in the camera view
   • Ensure nothing else in the frame resembles a hand

2. Low FPS:

   • Close other applications using the camera or CPU
   • Reduce camera resolution in the code if needed

3. MediaPipe installation errors:

   • Ensure you're using a compatible Python version (3.8-3.10 recommended)
   • Try installing MediaPipe with pip install mediapipe --no-deps
   • For Apple Silicon Macs, use Rosetta 2 terminal to install dependencies

You can modify the following parameters in the code:

• min_detection_confidence: Increase for better accuracy, decrease for better performance
• min_tracking_confidence: Adjust tracking stability vs. performance
• gesture_stable_threshold: Change how many consistent frames are required for stability
• Display colors and layout can be adjusted in the code

This project is licensed under the MIT License - see the LICENSE file for details.

• MediaPipe for the hand detection model
• OpenCV for computer vision functionality