Face Detection App (real-time)

What is this?

• Test repo for python app for face detection
• Libraries used:
  • DeepFace by Sefik Ilkin Serengil
  • MTCNN
  • OpenCV-Python
  • TensorFlow
  • MediaPipe

Note: create a virtual environment first, then install requirements:

• python -m venv ./venv
• pip install -r requirements.txt

DeepFace

To run, use app-df.py.

• Uses VGG-Face face detection model as default model.
  • Installs them on code run at C:\Users\<userfolder>\.deepface\weights
  • Database (db) folder is used for One-Shot Learning. multiple faces/expressions can be used as a baseline to compare to our model used.
    • It compares input img from video camera to our base pics in the db, with the help of our model

OpenCV - Bounding Box

To run, use app-cv.py.

• Simple face detection using Haar Cascade Classifier.
• Outputs face recognition bounding box, along with x, y coordinates of bounding box.

OpenCV - Face Landmarks

To run, use app-cv-2.py.

• Simple face detection same as app-cv, but with facial landmarks.
• Outputs face recognition bounding box, and points for facial landmarks.

MediaPipe

To run, use app-mp-land.py.

• Simple face detection with face mesh.
• Outputs face recognition face mesh.
• done!