Scribd
Upload
13 views4 pages

1 Face Recognition Intro and Overview

Uploaded by

drblood240
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
13 views4 pages

1 Face Recognition Intro and Overview

Uploaded by

drblood240
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 4

1

Python face-recognition Library Introduction and Overview

Python face-recognition library is a simple, user-friendly library with methods


useful for to recognize and manipulate faces from Python. More details can be
found from www.pypi.org/project/face-recognition/.

Python library Face-recognition is built on three important foundations.

1. CMake
2. Dlib
3. Open CV

CMake: CMake is a cross-platform free an open


source software tool. This is used to manage the

Copyright Dr. Xavier Chelladurai 17 July 2021


2

software building process using compiler


independent method.

1. Dlib is a dynamic library. This is actually a

modem C++ to solve real life problem. This


contains machine learning algorithms and tools
for building complex software in C++ to solve real
life problem. Most of the Machine Learning
packages are built on Dlib

2. Open CV (opensource computer vision)


This is a very popular opensource library
implementing Computer Vision algorithms using
Machine Learning.

Major Features

This library has the following key features:

1. Face detection and location

Copyright Dr. Xavier Chelladurai 17 July 2021


3

Face detection is about detecting human faces in a picture. This identifies all
the human faces in the picture and locate them in the 2-dimensional
representation as rectangle frames. This can be done using two different
methods.

1. Histogram Oriented Gradient approach denoted as hog approach.


2. Deep Learning based Convolution Neural Network (CNN) based
approach
While hog approach is simple and runs in less computing time its
accuracy is low. CNN is very complex and takes more time but has a
better accuracy level.

Model Speed Accuracy

HOG Fast Low

CNN Slow High

2. Face Encoding

Face Encoding is about extracting the features of faces and convert them into
their respective face signatures. The signature is farther used for comparing
faces and recognize a resemblance.

3. Face Resemblance Identification

Given two faces, it is possible to check the resemblance. This is done as


follows:

1. Load face f1 and f2


2. Find Face encoding of f1 and f2 and get their signatures s1 and s2.
3. Find the distance between s1 and s2 distance (s1, s2)
4. If distance (s1, s2) is small we say the face f1 and f2 have resemblance.

4. Face Landmark Identification

Copyright Dr. Xavier Chelladurai 17 July 2021


4

It can identify important landmarks on human faces. The following are


identified:

1. Chin
2. Left eyebrow
3. Right eyebrow
4. Nose bridge
5. Nose tip
6. Left eye
7. Right eye
8. Top lip
9. Bottom lip.

5. Face recognition

There are some known faces that are registered in the memory. When a new
face fu is received we can recognize this new face fu as one of the known faces.
This feature is used to used in the face-recognition-based Employee
Attendance System and Immigration Systems.

Copyright Dr. Xavier Chelladurai 17 July 2021

You might also like