Overview

This is an early api prototype made for an automated visitor/employee registration system using facial recognition which I had in mind. This api is built on top of Microsoft Face Api. It only requires a single api call for the following

• registering users
• identifying users
• deleting users

and also has the following additional features

• ensure user is only registered once in its own group (meaning Face Api only keeps a single person id for the user)
• detect if the person in the image has registered in a certain person group is specified. if person group is not specified, then it will check person against all person groups.

Use Cases

This api can be implemented over the various use cases:

• Registering people and authenticating them using their faces

How to deploy the api

Prerequisites:

• Microsoft Face Api Key
• Azure Table Storage Key
• Azure Table Storage Table

1) Create an external environment config file

Create a file config.ini and save it to folder regident/config/config.ini. Fill it with the contents below.

[DEFAULT]
FACEAPI_KEY = 
FACEAPI_BASEURL = https://<region>.api.cognitive.microsoft.com/face/v1.0
FACEAPI_REGION = 
AZURESTORAGE_KEY = 
AZURESTORAGE_CONNECTIONSTRING = 
AZURESTORAGE_ACCTNAME = 
AZURESTORAGE_TABLENAME = 

2) Set up the virtual environment

python -m venv venv
call venv/Scripts/activate.bat
pip install -r requirements.txt

3) Deploy the Flask Api to AWS using Zappa

del /S *.pyc
zappa deploy dev

The site is now live at the url printed in the console! Check out this article for more info on how to deploy:

https://medium.com/@houdinisparks/how-i-build-an-authenticated-serverless-flask-api-with-zappa-and-docker-for-a-model-582fc48fa0e0

How to use the api

Person

• Create person: POST /api/v1/person
  • Content-Type: multipart/form-data:
    • str:id
    • str:persongroupid
    • str:name
    • file:images
• Get person's details: GET /api/v1/person?id=<int:id>
• Add face to person: PUT /api/v1/person?id=<int:id>&persongroupid=<int:persongroupid>
  • Content-Type: multipart/form-data:
    • file:images
• Remove person: DELETE /api/v1/person?id=<int:id>&persongroupid=<int:persongroupid>

PersonGroup

• Create person group: POST /api/v1/persongroup
  • Content-Type: multipart/form-data:
    • str:persongroupid
    • str:name
    • str:data
• Get person group details: GET /api/v1/person?persongroupid=<int:persongroupid>
• Delete person group DELETE /api/v1/person?persongroupid=<int:persongroupid>

Identify

• Identify image for person: POST /api/v1/identify
  • Content-Type: multipart/form-data:
    • float:threshold
    • int:max_candidates_return
    • str:persongroups
    • file:image

Future Implementations:

• Save faces to azure blob storage
• Have more friendly http error codes. Now is just using the Azure http error codes.