Hand Gesture Recognition using OpenCV and MediaPipe

SkillCraft Technology - Machine Learning Internship

Task 04

Real-time hand gesture recognition from a live webcam feed using a dual-engine approach - MediaPipe Hands for landmark-based detection and OpenCV contour + convex hull as an automatic fallback.

---

Features

• Real-time webcam detection
• Multi-hand tracking (Left / Right identification)
• Finger counting (0 - 5)
• Named gesture recognition (9 gestures)
• 21-point landmark visualization
• Live FPS display
• Auto-fallback to OpenCV skin-segmentation when MediaPipe cannot detect a hand
• Gesture smoother (reduces flicker with rolling-window + hold timer)
• Interactive HUD overlay

---

Technologies Used

Tool	Purpose
Python 3.9+	Core language
OpenCV	Video capture, contour analysis, UI rendering
MediaPipe	Hand landmark detection (21 keypoints)
NumPy	Numerical operations and angle computation
Collections (deque)	Gesture smoothing buffer

---

Recognised Gestures

Gesture	Trigger Condition
Fist	All fingers curled
Open Hand	All 5 fingers extended
Thumbs Up	Only thumb up, tip above wrist
Peace	Index + middle extended
OK Sign	Thumb tip near index tip, rest extended
Pointing	Only index finger extended
Rock On	Index + pinky extended
Spider-Man	Thumb + index + pinky extended
Call Me	Thumb + pinky extended

---

Architecture

Webcam Frame
     |
     |---> MediaPipe Hands ---> Landmark Detection ---> Gesture Classification
     |           |                                              |
     |    (if no hand detected)                                |
     |           |                                             |
     +---> OpenCV Fallback ---> Skin Mask ---> Convex Hull ---> Finger Count
                                                               |
                                                       Gesture Smoother
                                                               |
                                                          HUD Overlay

---

How to Run

1. Clone the repository

git clone https://github.com/Sowmya-116/SCT_ML_4.git
cd SCT_ML_4

2. Install dependencies

pip install -r requirements.txt

3. Run the script

python hand_gesture_recognition.py

Press Q or ESC to quit.

---

Project Structure

SCT_ML_4/
|-- hand_gesture_recognition.py   # Main script
|-- requirements.txt              # Python dependencies
|-- README.md                     # Project documentation

---

Requirements

opencv-python>=4.8.0
mediapipe>=0.10.0
numpy>=1.24.0

---

Author

Sowmya - SkillCraft Technology ML Internship, Task 04

---

License

This project is for educational purposes under the SkillCraft Technology ML Internship program.