SSH Facenet Face Recogniton

Introduction

This repository includes a face recognition and detection system based on SSH face detection network and facenet feature extraction network.

This project can run on Ubuntu 16.04, using the python3.5 or python2.7 environment.The SSH network is built using caffe and the facenet network is built using Tensorflow.The two networks can be used combination or independently.

Contents

1. SSH Face Detection
2. Facenet Face Recognition
3. Open face recognition and detection remote service

SSH Face Detection

This project was modified based on the author's source code of the SSH network.

More details of the network can be found in this paper:"SSH: Single Stage Headless Face Detecto"

1. Install

(1) caffe-GPU and pycaffe are required

(2) Install python requirements:

cd SSH
pip install -r requirements.txt

(3) Run make in the SSH\lib directory:

cd lib
make

2. Running the demo

To run the demo, first, you need to download the provided pre-trained SSH model. Running the following script downloads the SSH model into its default directory path:

bash SSH/scripts/download_ssh_model.sh

Or you can download it from the Baidu cloud disk I provided.

By default, the model is saved into a folder named SSH/data/SSH_models .

After downloading the SSH model, you can run the demo with the default configuration as follows:

cd SSH
python demo.py

If everything goes well, the following detections should be saved as data/demo/demo_detections_SSH.png.

3.Training a model

If you want to train your own face detection model, you can see a detailed tutorial on this github

Facenet Face Recognition

More details of the network can be found in this paper:FaceNet: A Unified Embedding for Face Recognition and Clustering

1. Introduction

The face recognition system first uses SSH for face detection, then the cropped picture is input into the facrnet and encoded into a 128-dimensional vector, and then compared with the database, each two faces get a distance.The database face,which has the smallest distance and the distance is less than the preset threshold, is the recognition result.

2. Install

You can quickly start facenet with pretrained Keras model (trained by MS-Celeb-1M dataset).

-Download model from here and save it in model/keras/

You can also create Keras model from pretrained tensorflow model.

-Download model from here and save it in model/tf/ -Convert model for Keras in facenet\notebook\tf_to_keras.ipynb

3. Running the demo

(1) use SSHDetec Network.

first ，Open SSH face detection remote service

python SSH/service.py

and then Run the Jupyter Notebook SSH_facenet.ipynb in facenet\notebook
(2)Only run Facenet Network

-Run the Jupyter Notebook demo-images.ipynb in facenet\notebook\ in facenet\notebook\

3.Training a model

-If you want to train your own face detection model, you can see a detailed tutorial on this Wiki

-Convert model for Keras in facenet\notebook\tf_to_keras.ipynb

Open face recognition and detection remote service

(1)Open a command line window(ctrl+alt+T/Wins +r),and open the face detection service.

python SSH/service.py

(2)Open a command line window(ctrl+alt+T/Wins +r),and open the face recognition service.

python facenet/Verificationservice.py

(3)Use http protocol to access services.The demo is in the serviceTest\serviceTest.ipynb