Py68FaceZones

中文

Overview

Py68FaceZones uses dlib and OpenCV (cv2) to process face images, detect 68 facial landmarks, draw custom cheek-region Bezier paths, and export cropped cheek areas.

Features

• Automatically detects faces in an image.
• Draws all 68 dlib facial landmarks with index labels.
• Builds left and right cheek regions from facial landmarks using Bezier curves.
• Exports cropped cheek-region images for both sides of the face.

Requirements

• Python 3.12
• numpy<2
• opencv-python==4.8.1.78
• dlib-bin==19.24.2.post1
• imutils==0.5.4

Install the dependencies with:

python -m pip install -r requirements.txt

File Structure

• shape_predictor_68_face_landmarks.dat: dlib 68-point facial landmark predictor file.
• res/: Directory for input images.
• res/result/: Directory for generated output images.

Usage

1. Make sure all dependencies are installed and place shape_predictor_68_face_landmarks.dat in the project root.
2. Put the input image in the res/ directory.
3. Edit the file_name variable in main.py to specify the input filename without the extension. For example, file_name = "test" reads res/test.jpg.
4. Run the program:

python main.py

5. The generated images will be saved in res/result/.

Output Images

After execution, the following output files are generated in res/result/:

• test_landmarks.jpg: The original face image with all 68 landmark points and their indices drawn on top. This is useful for verifying that face detection and landmark localization are working correctly.
• test_full.jpg: The original image with green Bezier curve outlines showing the left and right cheek regions.
• test_left_zone.jpg: The cropped left cheek region defined by the custom path. Pixels outside the selected region are filled with white.
• test_right_zone.jpg: The cropped right cheek region defined by the custom path, using the same extraction logic as the left side.

Example Outputs

Landmark Detection

Cheek Region Paths

Extracted Left Cheek

Extracted Right Cheek

Error Handling

• If shape_predictor_68_face_landmarks.dat is missing, or the specified image does not exist in res/, the program will raise an error.
• If no face is detected in the image, the program will report an error.