manivannanmurugavel / pyfacy

This Package used for Face Recognition with Machine Algorithm

Installing Steps for requirements python package

Installing dlib on Ubuntu

The following instructions were gathered on Ubuntu 16.04 but should work on newer versions of Ubuntu as well.

To get started, let’s install our required dependencies:

sudo apt-get update
sudo apt-get install build-essential cmake
sudo apt-get install libopenblas-dev liblapack-dev
sudo apt-get install libx11-dev libgtk-3-dev
sudo apt-get install python python-dev python-pip
sudo apt-get install python3 python3-dev python3-pip

after

pip install dlib

Installing pyfacy models on Ubuntu

pip install pyfacy_dlib_models

Installing imutils on Ubuntu

pip install imutils

Installing numpy, scipy and sklearn

pip install numpy
pip install scipy
pip install scikit-learn

Installing pyfacy

pip install pyfacy

It's implemented with face encodings

Examples:

Read Image

from pyfacy import utils
img = utils.load_image('<image src>')
ex:
img = utils.load_image('manivannan.jpg')

Face Encodings:

from pyfacy import utils
img = utils.load_image('<image src>')
encodings = utils.img_to_encodings(img)

Compare Two faces

from pyfacy import utils
image1 = utils.load_image('<image1 src>')
image2 = utils.load_image('<image2 src>')
matching,distance = utils.compare_faces(image1,image2)

Note: The compare_faces return Boolean and Distance between two faces

Example for Face Recognition using ML

Implementing Algorithms

1. KNN - K-Nearest Neighbors
2. LOG_REG_BIN - Logistic Regression with two classes
3. LOG_REG_MUL - Logistic Regression with more than two classes
4. LDA - Linear Discriminant Analysis

Training , Save model and Predict Image

from pyfacy import face_recog
from pyfacy import utils
mdl = face_recog.Face_Recog_Algorithm()
# Train the Model
# Implemented only four algorithms above mentioned and put the shortform
mdl.train('pyfacy/Test_DS',alg='LOG_REG_MUL')
# Save the Model
mdl.save_model()
# Predicting Image
img = utils.load_image('<image src>')
mdl.predict(img)

Loading model and Predict Image

from pyfacy import face_recog
from pyfacy import utils
mdl = face_recog.Face_Recog_Algorithm()
# Load Model
mdl.load_model('model.pkl')
# Predicting Image
img = utils.load_image('<image src>')
mdl.predict(img)

Face Clustering

Cluster the image_src

from pyfacy import face_clust
# Create object for Cluster class with your source path(only contains jpg images)
mdl = face_clust.Face_Clust_Algorithm('./pyfacy/cluster')
# Load the faces to the algorithm
mdl.load_faces()
# Save the group of images to custom location(if the arg is empty store to current location)
mdl.save_faces('./pyfacy')

pyfacy's Issues

Clustering problem

Hi

When I try to cluster the pictures, but it cannot work.
when it loads pictures, some error happened, such as:

ImportError: [joblib] Attempting to do parallel computing without protecting your import on a system that does not support forking. To use parallel-computing in a script, you must protect your main loop using "if name == 'main'". Please see the joblib documentation on Parallel for more information

How to calculate the accuracy in terms of probability or percentage ?

Hi Manivannan,

I have a doubt regarding calculating the accuracy using the face_distance.
Example:
I want the convert the face_distance values to the percentage %.

Accuracy problem

Hi, i tried your face clustering example code:

from pyfacy import face_clust
# Create object for Cluster class with your source path(only contains jpg images)
mdl = face_clust.Face_Clust_Algorithm('./pyfacy/cluster')
# Load the faces to the algorithm
mdl.load_faces()
# Save the group of images to custom location(if the arg is empty store to current location)
mdl.save_faces('./pyfacy')

but the result is not that accurate, some different people were put into the same folder...

how to get better accuracy and how do I change face distance tolerance in large dataset?