/multi_factor_authentication

The biological MFA(multi-factor authentication) system ues InsightFace to extract face features and DeepSpeaker to extract voice features, and then performs hybrid authentication in Milvus.

Primary LanguagePython

❗❗ This repo will no longer be maintained, please visit https://github.com/milvus-io/bootcamp ❗ ❗

Biological multi-factor authentication system

The biological multi-factor authentication system presented in this paper uses InsightFace to extract the voice features and then does hybrid authentication in Milvus.

Preparation

Build System

Before setting up a biological multi-factor authentication system, start the Milvus service and deploy the system and build an image using Docker as described below.

Deploy with Docker

  • Run MFA-demo

$ docker run -td -p 5003:5000 -e API_URL=https://192.168.1.85:5003 -e "MILVUS_HOST=192.168.1.85" -e "MILVUS_PORT=19530" -e "PG_HOST=192.168.1.85" -e "PG_PORT=5432" milvusbootcamp/mfa-demo:0.2.0

Description of parameters related to the above start command.

Parameter Description
-p 5003:5000 -p represents the port mapping between the host and the image
-e API_URL=https://192.168.1.85:5003 -e represents the system parameter mapping between the host and the image
Change 192.168.1.85 to the IP address of the server where MFA-demo is currently started, and 5003 is the port mapped to the host.
-e "MILVUS_HOST=192.168.1.85" Please change 192.168.1.85 to the IP address of the server where Milvus docker starts.
-e "MILVUS_PORT=19530" Please change 19530 to the server port number to start Milvus docker.
-e "PG_HOST=192.168.1.85" Please change 192.168.1.85 to the IP address of the server where Postgres is started.
-e "PG_PORT=5432" Please change 5432 to the server port for starting Postgres.

Build mirror deployments

# Pull mfa code 
$ git clone https://github.com/milvus-io/bootcamp.git
$ cd bootcamp/solutions/MFA/webserver
# Build mfa-demo image
$ docker build -t mfa-demo:0.2.0 .
# Run mfa-demo,for startup commands, see "Deploying with Docker".
$ docker run -td -p 5003:5000 -e API_URL=https://192.168.1.85:5003 -e "MILVUS_HOST=192.168.1.85" -e "MILVUS_PORT=19533" -e "PG_HOST=192.168.1.85" -e "PG_PORT=5432" mfa-demo:0.2.0

Note：To build the image, you need to download the face_embedding model, download link: https://pan.baidu.com/s/18EWcP5YJmeDrY1A8_k09pw , Extraction code: 82ht; to download the deep speaker model, download link: [https://pan. baidu.com/s/16_moPcoUGah1dqdDtEQreQ](https://pan. baidu.com/s/16_moPcoUGah1dqdDtEQreQ), Retrieval Code: 11vv.

After Downloading, unzip mfa-models，and move the models folder to bootcamp/solutions/MFA/webserver/face-embedding ; move ResCNN_triplet_training_checkpoint_265.h5 to bootcamp/solutions/MFA/webserver/src/deep_speaker/checkpoints.

System Usage

Type https://192.168.1.85:5003 (the API_URL specified by the launch MFA-demo) into your mobile or client browser (Google Chrome is recommended) to start the biological multi-factor authentication.

  • New User Registration

    First, click New User to enter the information in the system.

​ Then fill in the box with your nickname, e.g. Milvus, and click on Click to Record to record the video, if the browser pops up and you need access to the camera and microphone, select Yes, then you will record about 5 seconds of video in the system.

​ The following GUI will appear after a successful system entry.

  • System authentication

    Clicking Click To Identify will authenticate and also record the video in the system for about 5s.

​ Successful system authentication is shown below, with the image showing the verifier's face. The Milvus logo has been replaced here.