Face Mesh Landmark Extraction

This project uses the "MediaPipe Face Mesh solution" to detect facial landmarks in a single image and extract their x, y, and z coordinates. The coordinates are written to a .txt file, and the image is saved with landmarks overlaid on the face.

Table of Contents

• [Introduction]
• [Features]
• [Requirements]
• [Installation]
• [Usage]
• [Output]
• [Project Structure]

Introduction This project detects facial landmarks from an input image using MediaPipe's Face Mesh. It extracts the x, y pixel coordinates, and the z depth coordinate for each of the 468 landmarks and saves them to a .txt file. It also generates an image with the detected landmarks drawn over the face.

Features

• Detects 468 face mesh landmarks using "MediaPipe".
• Writes the landmark coordinates (x, y in pixels, z as relative depth) to a text file.
• Saves the processed image with the landmarks drawn as small circles.
• Processes a single image with customizable input/output paths.

Requirements

• Python 3.x
• OpenCV (opencv-python)
• MediaPipe

Installation

1. Clone the Repository:

   git clone https://github.com/yourusername/face-mesh-landmarks.git
   cd face-mesh-landmarks

2. Install the Required Libraries: Install the dependencies using pip:

   pip install opencv-python mediapipe

Usage

1. Set the Image Path:
   Update the image_path variable in the script with the location of your image. The default path is set to 'D:/images/image_1.jpg'.

2. Run the Script:
   Execute the Python script to extract face mesh landmarks and save the output.

   python face_mesh_landmarks.py

3. View the Results:
   The script will output two files:

   • A .txt file with the x, y, z coordinates of the landmarks.
   • An image file with landmarks drawn on the original image.

Example:

# Define the path of the single image
image_path = 'D:/images/image_1.jpg'

Once you run the script, it will output:

• landmarks_output.txt: This contains the landmark coordinates.
• output_landmarks_image_1.jpg: The image with face landmarks.

Output

1. Landmark Coordinates File (landmarks_output.txt) This file will contain the following columns:

• IMAGE_NAME: The name of the image processed.
• LANDMARK_INDEX: The index number of the landmark (0-467).
• X_PIXEL: The x-coordinate of the landmark in pixel values.
• Y_PIXEL: The y-coordinate of the landmark in pixel values.
• Z_COORDINATE: The z-coordinate, which is a relative depth value.

Sample output:

IMAGE_NAME      LANDMARK_INDEX   X_PIXEL    Y_PIXEL    Z_COORDINATE
image_1.jpg     0                500        600        0.003402
image_1.jpg     1                520        580        0.003600
...

2. Output Image An image with the face mesh landmarks drawn as green circles will be saved with the filename output_landmarks_<image_name>.jpg.

Project Structure

face-mesh-landmarks/
│
├── face_mesh_landmarks.py    # Main Python script to run the project
├── README.md                 # Project documentation
├── landmarks_output.txt      # Generated output file containing the landmarks (created after running the code)
├── output_landmarks_image_1.jpg  # Image with landmarks (created after running the code)
└── images/
      └── image_1.jpg         # Example input image (replace with your own image)