DEV Community

Cover image for Hello administrator, I can see you
Paula
Paula

Posted on

Hello administrator, I can see you

Not long ago I published a post about facial recognition and how to evade it, because I was working with it due to a CTF. I'm playing with the same library in order to create a 2nd login page that requires a picture of the administrator. I wanted a simple POC, so I used a simple Flask structure for it. First we need a basic login setup in python, for regular application you should probably use a decent DB managing, you guys know more about that. For know let's see the scratch, first of all we are going to need a lot of things to make this work:

#app.py

from flask import Flask
from flask import Flask, flash, redirect, render_template, request, session, abort
from flask import redirect

import face_recognition
from PIL import Image, ImageDraw
import numpy as np

import os
...
Enter fullscreen mode Exit fullscreen mode

Now let's define routes

@app.route('/')
def home():
    if not session.get('logged_in'):
        return render_template('login.html')
    else:
        #this is a dummy page for my trials :)
        return render_template('dashboard.html')

@app.route('/login', methods=['POST'])
def do_admin_login():
    #me: I work in security. 
    #also me: 
    if request.form['password'] == 'password' and request.form['username'] == 'user':
        #this is going to be our first change, 
        #we are using an image-upload page in between 
        return render_template('showmethatprettyface.html')
    else:
        return home()

Enter fullscreen mode Exit fullscreen mode

In showmethatprettyface.html we are using a simple image upload setup

<div class="container">
  <div class="row">
    <div class="col">

      <h1>Upload an image</h1>
      <hr>

      <form action="/showmethatprettyface" method="POST" enctype="multipart/form-data">

        <div class="form-group">
          <label>Select image</label>
          <div class="custom-file">
            <input type="file" class="custom-file-input" name="image" id="image">
            <label class="custom-file-label" for="image">Select image...</label>
          </div>
        </div>

        <button type="submit" class="btn btn-primary">Upload</button>

      </form>

    </div>
  </div>
</div>
Enter fullscreen mode Exit fullscreen mode

This way back in our app.py we can retrieve the uploaded picture using a new route

@app.route("/showmethatprettyface", methods=["GET", "POST"])
def upload_image():

    if request.method == "POST":

        if request.files:

            picture = request.files["image"]
Enter fullscreen mode Exit fullscreen mode

Now, the funny part is using this picture for face recognition. For that, we need known pictures, first.

import stuff blah blah
...

# Load a second sample picture and learn how to recognize it.
salita_image = face_recognition.load_image_file("salita.jpg")
salita_face_encoding = face_recognition.face_encodings(salita_image)[0]

# Create arrays of known face encodings and their names
known_face_encodings = [
    salita_face_encoding
]
known_face_names = [
    "Salita"
]
Enter fullscreen mode Exit fullscreen mode

Salita's gonna be our known user. Now again in our route we do the rest of the logic which, in fact, is explained here.

@app.route("/showmethatprettyface", methods=["GET", "POST"])
...

            # Load an image with an unknown face
            unknown_image = face_recognition.load_image_file(foto)

            # Find all the faces and face encodings in the unknown image
            face_locations = face_recognition.face_locations(unknown_image)
            face_encodings = face_recognition.face_encodings(unknown_image, face_locations)

            # Convert the image to a PIL-format image 
            pil_image = Image.fromarray(unknown_image)

            # Create a Pillow ImageDraw Draw instance to draw with
            draw = ImageDraw.Draw(pil_image)

            # Loop through each face found in the unknown image
            for (top, right, bottom, left), face_encoding in zip(face_locations, face_encodings):
                # See if the face is a match for the known face(s)
                matches = face_recognition.compare_faces(known_face_encodings, face_encoding)


                if True in matches:
                    return render_template('successfulldummypage.html')
                else:
                    return render_template('booodummypage.html')

Enter fullscreen mode Exit fullscreen mode

So if in the picture the user uploads in the page appears Salita the successfulldummypage.html will load and either way booodummypage.html. If I was Salita I would be very careful of the pictures of me in social networks. Anyway very super disclaimer: This is for fun, don't trust it for serious business! also pretty insecure!

It's a lot of fun to try, tho.

Top comments (0)