Real-Time AI-based Virtual Mask Filter 🎭

An AI-powered application that overlays virtual masks onto users' faces in real time. The app first detect your face on input video with a bounding box and keypoints. Then, a chosen mask (e.g., anonymous, squidgame man) will be applied on your face

📌 Features

• Real-time face detection and landmark recognition.
• Applies virtual masks with precise alignment.
• Works efficiently on both desktop and mobile devices.
• Optimized for performance using deep learning techniques.

🔍 How It Works

The application follows a structured pipeline to process and enhance facial images effectively:

1. Face Detection: Leverages both OpenCV’s Deep Learning-based Face Detector and MediaPipe Face Detection for accurate and efficient face recognition in images and videos.
2. Facial Landmark Detection: Uses a pre-trained EfficientNet model with transfer learning to detect 68 facial landmarks.
3. Face Alignment: Employs the Delaunay Triangulation algorithm to align facial features for precise transformations.
4. Filter Application: Once aligned, various filters—such as beauty enhancements, cartoon effects, or virtual masks—can be applied dynamically.

🛠 Key Technologies Used

• Python
• PyTorch - for deep learning models
• OpenCV - for image processing
• EfficientNetB0 - for facial landmark detection
• Delaunay Triangulation - for accurate mask mapping

🚀 Installation

Follow these steps to set up the project:

# Clone the repository
git clone https://github.com/Lhhiep-maxcode/Filter.git
cd Filter

# Create a virtual environment (optional but recommended)
python -m venv env
source env/bin/activate  # On Windows use: env\Scripts\activate

# Install dependencies
pip install -r requirements.txt

🌟 Running the Application

After setting up the environment, open the Face detection.ipynb file and execute all the cells. Once completed, you’ll see impressive results as demonstrated in the introduction. 🚀