Face Recognition with DeepFace and OpenCV

Overview

This script performs real-time face recognition using OpenCV and DeepFace. It captures video from your webcam, compares the detected face to a reference image, and displays appropriate messages based on whether the face is found and if it matches the reference.

Dependencies

• DeepFace for facial recognition
• OpenCV for video capture and image processing

Setup Instructions

1. Install Anaconda (if not already installed)

   • Download and install Anaconda for Python 3.10.

2. Create a Conda Environment Using environment.yml

   • Clone or download this project to your local machine.
   • In the project directory, there is an environment.yml file, which contains all the necessary dependencies for the script.
   • Create the Conda environment by running:

     conda env create -f environment.yml

3. Activate the Environment

   • Once the environment is created, activate it:

     conda activate FaceRec

4. Install Additional Dependencies (if needed)

   • In case DeepFace or OpenCV is missing, you can manually install the necessary dependencies with:

     pip install deepface opencv-python

5. Prepare Reference Image

   • Place the reference image (the one you want to compare against) in your project directory and update the reference_img path in the code:

     reference_img = cv2.imread("path_to_your_image.jpg")

6. Run the Script

   • Run the script:

     python face_recognition.py

   • The webcam will start capturing, and the script will analyze each frame to see if the face matches the reference.

How it Works

1. Video Capture:

   • The script captures frames from your webcam using OpenCV (cv2.VideoCapture(0)).

2. Face Detection and Matching:

   • Every 30 frames, the check_face function is called in a separate thread. This function:
     • Uses DeepFace's verify function to compare the captured face with the reference image.
     • Updates the flags (face_match, face_found) depending on the result.

3. Display Messages:

   • The script overlays messages on the video feed based on the face matching result:
     • "Hi, {Name}" if a matching face is detected.
     • "Face found no match!" if a face is found but doesn't match.
     • "Face not found!" if no face is detected.

4. Exit:

   • Press q to exit the video feed.

Important Notes

• Ensure that your webcam is properly set up and recognized by your system.
• The script uses threading to handle the face comparison without blocking the main video loop.
• You may need to adjust the reference image path based on your system and file structure.

Resources

• DeepFace Documentation
• OpenCV Documentation