FACIAL RECOGNITION IN SOFTWARE FOR SECRETS MANAGEMENT

Abstract - Facial recognition is a type of advanced authentication method, which relies on biometric characteristics of a user’s face to confirm their identity. In the context of software for secrets management, this method has proven to be an effective alternative to traditional methods, such as passwords, providing greater security and ease of use. This paper researches the process of creating a machine model for facial recognition and how that model can be integrated into software for secrets management. The goal is to create a web application through which secrets can be safely accessed after facial verification; the model is successfully integrated if the application allows access only to the desired user.

The model can be used for one shot verification in a Jupyter environment

Or integrated with an application, in this case with Django and managed through the integrated admin panel

@require_POST
def process_image(request):
    img = request.body.decode('utf-8')
    img = img.split(',', maxsplit=1)[1] # Remove base64 header
    img = np.frombuffer(base64.b64decode(img), np.uint8)
    img = cv2.imdecode(img, cv2.IMREAD_COLOR)

    # Process the image as needed
    img = preprocess_np(cut_frame(img))
    results, verified = verify(img, siamese_model, APP_PATH, 0.7, 0.6)

    if verified:
        request.session['verified'] = True

    # Return the result
    return JsonResponse({'verified': verified})

The full code for the Django application implementing this work can be found here: NewPass