web UI for who's that pokemon project

Disclaimer 1: This webapp has the only purpose of showing the potential of the models explored inside the Research block of the project. This is not intended to be a real world application.

Disclaimer 2: Don't trust the username & pass login, it is just a demonstrative login with default username "admin" and password "admin". Nothing more than a graphical feature to simulate a system administrator that adds new people to the database.

• A system administrator can login using Id and Password in order to manage the registered users or register a new user
  • To register a new user a picture of his face must be selected
  • Scrolling down in the administrator panel is possible to manage the current registerd users
  • Disclaimer: Don't trust the username & pass login, it is just a demonstrative login with default username "admin" and password "admin". Nothing more than a graphical feature to simulate a system administrator that adds new people to the database.
• The face login can be activated clicking on the face icon on the right
  • The webcam will be activated
  • If face crop and align option is ticked then the login probability is higher. The face will be cropped using the image transformation described in the research block of the project.
  • If the liveness detection option is ticked then during the login the user must follow the steps shown below
  • If the login fails then the reason will be showed
  • Otherwise the name of the recognized person will be shown
    

Install the requirements with

pip install -r requirements

If you're not using the default url or port for the 'image-registration-service' please update the environment variable accordignly with:

export K8_IMAGE_REGISTRATION_SERVICE='http://<new url to the image registration service>:<port>'

• run the UI with python entry.py. This will start the web server at localhost:5006.
  
• In order to upload a new image to the database or use the login you must also run the ìmage-registration-service on the default 5005 port.
  
• If you plan to use the liveness detection the you have to run also the liveness-detection-service on the default 5007 port.
  

In order to deploy this service on Openshift the following config files must be updated

• The build configuration YAML file must be updated accordingly

  strategy:
      type: Source
      sourceStrategy:
        from:
          kind: ImageStreamTag
          namespace: openshift
          name: 'python:3.6'
        env:
          - name: K8_IMAGE_REGISTRATION_SERVICE
            value: 'http://<cluster address of the image registration service>:8080'
  

• Inside the service configuration YAML file update the route target port for the 8080-tcp spec

  spec:
    ports:
      - name: 8080-tcp
        protocol: TCP
        port: 8080
        targetPort: 5005
  

• Since this service must be exposed, be sure to setup the route.
• Force encryption over TLS for the route updating the route configuration YAML file accordingly

  annotations:
      app.openshift.io/connects-to: image-registration-service
      openshift.io/host.generated: 'true'
      haproxy.router.openshift.io/hsts_header: max-age=31536000;includeSubDomains;preload
  
  spec:
    tls:
      termination: edge