README.md

Face recognizer

Project to learn more about Computer Vision, specifically recognition and AI training. Recognizes faces from a given dataset.

To do list

• Emotion recognizer (medium)
• Lip reading (hard)

Usage

First step is to install all the depedencies from requirements.txt using:

pip install -r requirements.txt

Then, you should train the model on pictures you uploaded in the folder training using:

detector.py --train

Then, you can validate your model on all images in folder validation using:

detector.py --validate

Or a specific image using:

detector.py --image [path_image]

Finally, you can also use your model on videos:

detector.py --video [path_video]

Or on your webcam:

detector.py

The folder tree should look like this:

..
└── face_recognizer/
    ├── code/
    │   └── detector.py
    ├── training/
    │   ├── person0/
    │   │   ├── image_person00
    │   │   ├── image_person01
    │   │   └── ...
    │   ├── person1/
    │   │   ├── image_person10
    │   │   ├── image_person11
    │   │   └── ...
    │   ├── person2/
    │   │   ├── image_person20
    │   │   ├── image_person21
    │   │   └── ...
    │   └── ...
    ├── validation/
    │   ├── image0
    │   ├── image1
    │   └── ...
    ├── output/
    │   └── encodings.pickle
    └── ...  

usage: detector.py [-h] [--video VIDEO] [--image IMAGE] [--model MODEL] [--encodings ENCODINGS] [--train] [--validate]

optional arguments:
  -h, --help            show this help message and exit
  --video VIDEO         Path to the video file
  --image IMAGE         Path to the image file
  --model MODEL         Model to use for face detection (hog or cnn), default=hog
  --encodings ENCODINGS
                        Path to the encodings file (.pickle)
  --train               Train the model (from the folder 'training')
  --validate            Validate the model (from the folder 'validation')

Note: press 'q' to exit video or image and press 'space' to pause the video