michaelmano / facial-recognition

Facial Recognition

Setup

pip3 install sklearn Pillow face_recognition to get all required libraries for this project.

Training

Create a folder for each person using their name in lowercase and underscores for spaces e.g. jack_doe john_doe. Fill the folder with images of just that person and make sure no other faces are in the image and there is more than 1 image.

run python3 ./status_training.py, once complete the database.clf will be updated which will be used for the predictions.

Predictions

To run a facial recognition on an image all you have to do is run python3 ./status_prediction.py ./path/to/image.jpg and the script will return the results of anyone it picks up in the image, Results will be returned as JSON.

[
  {
    "name": "unknown",
    "coordinates": [982, 957, 1025, 914],
    "score": 0.6941232641733543
  },
  {
    "name": "john_doe",
    "coordinates": [660, 1447, 703, 1404],
    "score": 0.36921533611616386
  },
  {
    "name": "jack_doe",
    "coordinates": [391, 957, 434, 914],
    "score": 0.36302260895814914
  },
  {
    "name": "unknown",
    "coordinates": [430, 1567, 473, 1524],
    "score": 0.6223868744925297
  }
]

You can run the script on the test data provided like so python3 ./status_prediction.py test_data/faces.jpg

if no faces are detected an error will be returned {"error": "No faces found"}