Face Key Points Extractor

Overview

This project is a Python application designed for image processing with a focus on face feature extraction. It uses libraries like dlib and mediapipe for advanced face detection and processing.

---

Installation

1. Clone the repository.

2. Navigate to the project directory.

3. Create a virtual environment:

python -m venv .venv

4. Activate the virtual environment:

• On Windows:

.venv\Scripts\activate

• On Unix or MacOS:

source .venv/bin/activate

5. Install the dependencies:

pip install -r requirements.txt

---

Dlib Model Configuration

Before running the application, download the shape_predictor_68_face_landmarks.dat file for dlib:

1. Download the model file from Dlib's model repository or a trusted source.

2. Unzip the file and place it in the models directory at the root of the project.

3. Ensure the config.py in the settings directory is set correctly:

APP_PATH_PARAMS_JSON_FILE = 'params.json'
APP_PATH_MODEL_DLIB_FILE = 'models/shape_predictor_68_face_landmarks.dat'

---

Configuration parameters file

Customize the application behavior by modifying the params.json file in the root directory:

{
    "input_path": "",
    "models": [
        "dlib",
        "mediapipe"
    ],
    "images_format": [
        ".png",
        ".jpg",
        ".jpeg"
    ],
    "keypoints_color": [255, 255, 255],
    "margin_ratio": 0.1,
    "num_threads": 8
}

• input_path: Path to the directory containing the images to be processed.

• models: List of models to use for face feature extraction. Supported models are dlib and mediapipe.

• images_format: List of image file formats to be processed. For example: .png, .jpg, .jpeg.

• keypoints_color: Color of the keypoints (in BGR format) to be drawn on the images. It's an array of three integers representing Blue, Green, and Red color values.

• margin_ratio: The margin ratio around the detected face for cropping. A smaller value results in a tighter crop around the face.

• num_threads: The number of threads to use for image processing. Increasing this number can speed up processing on multi-core systems.

---

Usage

To run the application, use the following command:

python run.py [--debug]

The --debug argument is optional and enables the debug mode for additional logging and process details.

Key Components

• run.py: The entry point of the application. It parses command-line arguments and initializes the application with the given parameters.

• main.py: Contains the main logic for processing images using multithreading.

• extractor.py: Includes functions for face feature extraction using dlib and mediapipe.

• utils.py: Provides utility functions used across the application.

• config.py: Contains configuration settings for the application.

---

License

This project is open-sourced and available to everyone under the MIT License.