Facial Recognition Server

Introduction

This is the Rooster server-side code for facial recogntion and verification. This code will be deployed onto a cloud based server. This server is designed to provide efficient real-time matching capabilities. This server is designed to receive multiple image frames from a security camera and process those frames by looking for faces that match those in the Rooster database.

Prerequisites

Before running this server, ensure you have the following dependencies installed:

• Python
• Required Python packages (listed in requirements.txt)

Getting Started

1. Clone this repository to your local machine:

   git config credential.helper store
   git clone https://github.com/Rooster-AI/FaceMatchServer.git
   

Running Unit Tests

(python-env) -m unittest discover -s test

Using Docker

1. Build the docker image

   • from the Rooster/FaceMatchServer dir run

   docker build -t rooster-server .
   

2. (opt. 1)Run the docker image

    docker run -d -p 5000:5000 --restart always rooster-server

3. (opt. 2)Run a script that will restart docker image and run git pull nightly at 3am (set timezone - sudo timedatectl set-timezone America/Denver) a. make restart_docker.sh executable

   chmod +x restart_docker.sh

   b. add to crontab

   crontab -e

   c. add a new line to that file and save file

   0 3 * * * /path/to/restart_docker.sh

   d. verify scheduled task

   crontab -l

API

POST: /upload-images

• headers: application/json
• body:

{
   "images":["",""],
   "device_id": "string"
}

POST: /add-banned-person

• headers: application/json
• body:

{
   "full_name": "", 
   "drivers_license": "",
   "est_value_stolen": "", 
   "reporting_store_id": "",
   "is_private": "", 
   "description": "",
   "image_encoding":""
}

GET: /latest_database?model=ArcFace&backend=mtcnn

• URL params are optional, those are the defaults
• If the server has not compiled with that model and backend, will return an errror
• Returns:
  • a .pkl file as as attachment containing the latest encoded database of faces

API TODO

• /upload-images add store id to get the raspberry pi it was coming from