FaceRecognitionAPI

A Django OpenCV wrapper that acts as a RESTful API for face recognition

Dependencies

Make sure you have all the dependencies installed before running the API server. This project is dependent on:

• Python3 (brew install python3)
• OpenCV3 (brew install opencv3 --with-contrib --with-python3 --without-python) built with extra modules
• Django (pip install django)
• PIL (pip install Pillow)

If you encounter any problems installing OpenCV3 with Python3, consult this StackOverflow answer.

Installation

Download or clone the project and execute
python manage.py migrate
Once migrations finish executing, run the project using
python manage.py runserver

Usage

Requests

You can use the API by making a GET request to the following endpoints:

Endpoint	Description
/new	Used to add a user profile before training
/train	Used to add a known photo to a user profile and use it to identify the user later
/recognize	Given a photo, recognize the user in the photo
/users	Lists all registered users

/new

Creates a new user profile and prepares an internal folder to host all photos of this user's face.

/recognize

Recognizes a user based on a provided photo. Call should provide exactly one of the following parameters.

Parameter	Description	Type
imageBase64	Base64 encoded photo of the user to be recognized	string (Base64 encoded)
url	URL to the photo of the user to be recognized	string (Valid URL)

Example:

Example Request:

GET localhost:8000/recognize/?url=http://..

Example Response:

{
  "detected": true,
  "identity": 1,
  "user": {
    "first_name": "John",
    "last_name": "Appleseed",
    "username": "appleseedj",
    "email": "[email protected]",
    "id": 1
  },
  "box": [[246, 104, 948, 806]],
  "smiling": true
}

/train

Trains a registered user's model using a given photo of the user. The image is cropped, saved and used in the model on the next restart of the API server.

/users

Lists all users currently registered.

Take a look at the MirrorOS Project for an example of usage of the API.