🎭 Gesture-Controlled Image Filters using OpenCV & MediaPipe 🎨

📌 Overview

This project implements real-time hand gesture recognition to apply different image filters using a webcam.
The system detects six different hand gestures and applies corresponding filters to the webcam feed.

Supported Gestures & Filters:

Gesture	Effect
✋ Open Palm	No filter (Default)
✊ Fist	Grayscale
✌️ Peace Sign	Sepia
👍 Thumbs Up	Blur
☝️ Pointing Finger	Edge Detection
👌 OK Sign	Cartoon Effect

---

🔧 Installation

Ensure you have Python installed (>=3.7). Then, install the required dependencies:

pip install opencv-python mediapipe numpy torch torchvision scikit-learn pickle-mixin

Alternatively, install from the requirements.txt file:

pip install -r requirements.txt

---

🚀 Project Structure

gesture-controlled-filters/
│— collect_data.py       # Collects hand keypoints for model training
│— train_model.py        # Trains a gesture classification model
│— apply_filters.py      # Runs real-time gesture detection & filtering
│— gesture_data.pkl      # Saved dataset of hand keypoints (generated by collect_data.py)
│— gesture_model.pth     # Trained PyTorch model (generated by train_model.py)
│— requirements.txt      # Required dependencies
│— README.md             # Project documentation

---

🎥 How to Use

1️⃣ Collect Gesture Data

Run this script to collect 200 frames per gesture using MediaPipe Hands:

python collect_data.py

🛠 Instructions:

• Press Enter to start collecting for each gesture.
• Hold the gesture steady and slightly vary angles.
• Captures 200 samples per gesture.
• Press 'q' to quit data collection.

2️⃣ Train the Gesture Classification Model

Once data is collected, train the MLP neural network:

python train_model.py

💪 This will generate gesture_model.pth, which is the trained model.

3️⃣ Run Real-Time Gesture Recognition & Filtering

python apply_filters.py

👀 The webcam will display both the original frame and the filtered frame.
Try different gestures to see filters change in real time!

---

🛠 Troubleshooting

❌ No webcam feed appears

• Ensure your camera is enabled in system settings.
• Try running with cv2.VideoCapture(1) if using an external webcam.

❌ No gestures detected

• Increase min_detection_confidence in collect_data.py:

  hands = mp.solutions.hands.Hands(min_detection_confidence=0.3)

• Ensure good lighting and clear background.

❌ Filters not applying correctly

• Print model predictions:

  print(f"Detected gesture: {gesture}")

• If misclassification occurs, collect more data and retrain the model.

❌ Cartoon filter is not showing correctly

• Adjust cv2.adaptiveThreshold() in cartoon():

  edges = cv2.adaptiveThreshold(gray, 255, cv2.ADAPTIVE_THRESH_MEAN_C,
                                cv2.THRESH_BINARY, blockSize=9, C=9)

---

⭐ Contributing

Feel free to fork this repo and submit pull requests! Suggestions and improvements are always welcome.

🚀 Happy Coding! 🎬✨