opticrd / facial-auth-api

An API to verify users using their faces and document ID (Cédula) photo.

Facial Authentication is an API for verifying the user identity through the liveness detection process and comparing the Document ID photo with their faces.

Face Authentication API

Getting started

To authenticate a user you need to make a request to GET /verify sending the cedula and a short video (the video should be not more then 5 seconds) of the face of the user:

curl -i -H 'Content-Type: application/json' -d="cedula=${CEDULA_NUMBER}&source=data:video/webm;base64,${VIDEO_BASE64}"  http://{ENDPOINT}/verify

Also you can open a web demo going to https://{ENDPOINT}/:

Install

environment variables

# Optional env
PORT=80
HOST="0.0.0.0"
REDIS_HOST='localhost'
REDIS_PORT=6379
LOG_DIR='/var/logs/facial-auth-api/'

# Require
CEDULA_API_KEY=...
AWS_ACCESS_KEY_ID=...
AWS_SECRET_ACCESS_KEY=...
AWS_DEFAULT_REGION=...

To install it using docker with docker-compose:

git clone [e-mail]:opticrd/facial-auth-api.git
cd facial-auth-api
docker-compose up --build

To install it manually (python 3.6+ and pip should be installed):

# Deploy redis for json support
 docker run -p 6379:6379 -d --name redis-stack redis/redis-stack:latest

#
git clone [e-mail]:opticrd/facial-auth-api.git
cd facial-auth-api
pip install -r requirements.txt
uvicorn src.api:app

Socket.io Docs

Event	Description	Emitter	Input schema	Input data type	Output schema	Output data type	Error schema
verify	Event to verify users	False	{cedula: String, source: String, id: String}	json	Not output	N/A	N/A
result	Result of verification	True	Not input	N/A	{verified: Bool, face_verified: Bool, is_alive: Bool}	json	{error: String}

Authors

• Eddy Decena