Download as pdf or txt
Download as pdf or txt
You are on page 1of 3

1.

Good morning everyone. First, i would like to welcome and thank you for being here today to
witness our presentation. And second, i want to give a special thanks to my professors Mr. Khalid
minaoui and Miss hanae belmajdoub for assisting us all the way and who i’ve learned so much
from.

2.
Moreover, today we’re here to present to you a very important subject. It is the future f world
technologie, Deep learning. This last is a subset of machine learning, which in turn is a subset of AI.
Ai which stand for Artificial intelligence is a technique to make a machine mimic human be havior.
Machine learning is technique to achieve AI through algorithms trained with data. And finally, Deep
Learning is a type of machine learning inspired by the structure of the humain brain, which called
an artificial neural network.

3.
Now to understand the diffrence between deep learning and machine learning, we’ll make an
example of a machine that can diffrentiate between car pictures from non car pictures. Done with
machine learning, we’d have to tell the machine the features based on which these cars can be
diffrentiated, it could be the size or the shape. But in the other hand, with deep learning. The
features are picked out by the neural network without human intervention.

4.
As the picture shows, neural networks in general are formed by three types of layers. And each
layer is formed by many neurons. The first layer represent the input layer, it is where we input the
data to be processed. The second layer is called hidden layer. The number of these last ones depends
in the functionality of the network. And it is the responsible for pocessing the data that comes
thorght the input layer. And finally the output layer, yoo simply it is where we get the results.

5.
There are many types of neural networks. And you can see any of them in te pictures but the one
we’ll focus on is the one that’s called CNN.

6.
CNN which is a short for Convolutinal Neural Network, is a neural network that mostly used for
analizing IMAGES. And as all neural networks, CNN has it’s hidden layers.

7.
the first hidden layer of CNN is the convolutional layer. In this layer we aply filters on the input in
order to extract the features.

8.
the second layer is the max pooling layer. And in this one we hold on to the big features and
eliminate the small ones.

9.
the third layer is the flattening layer. And as the name suggest, in this later we turn the 3D image
matrix to 1D.

10.
and last but not least, the fully connected layer. This is where the image get classified based on the
features extracted from it.
11.
Based on what we learned about deep learning and CNN, we managed to put together a project
about class attendance management based on facial recognition. And as all projects in the
computing industry, ours has the hardware side and the software side. Starting with the Hardware,
i’ll let my colleague present it for you.

12.
Raspberry Pi is a small single board computer. By connecting peripherals like Keyboard, mouse,
display to the Raspberry Pi, it will act as a mini personal computer. Raspberry Pi is popularly used
for real time Image/Video Processing, IoT based applications and Robotics applications.

13.
The Raspberry Pi Camera Module v2 is a high quality 8 megapixel camera based around the Sony
IMX219 image sensor - allowing you to create HD video and still photographs.

14.
The programming language used in our project is Python.
Python is a popular programming language. It was released in 1991. it is used in many fields. Such
as, web development, software development, mathematics, system scripting.

15.
since our project mainly require dealing with pictures and videos, we’ve used cv2.
OpenCV or cv2 is a library of Python bindings designed to solve computer vision problems

16.
next we’ve used Face-recognition, which is a python library contains algorithms and pre-defined
function designed only for facial recognition.

17.
The pickle module implements binary protocols for serializing and de-serializing a Python object
structure. “Pickling” is the process whereby a Python object hierarchy is converted into a byte
stream, and “unpickling” is the inverse operation

18.
And to manage our database of informations. we’ve used sqlite, which is a C-language library that
implements a small, fast, self-contained, high relibiliy, full featured, SQL database engine.
And now to the last part of our presentation, which will be presented by my colleage

19.
when the program starts, it connects to the pictures database. This last one contains folders, and
each folder is named after one person and contains pictures of that person. The program circles
around every pictures for features extraction, and recognizes the person from the folder’s name.
This operation is what’s called training.
After the training is done, the results will be saved in a folder with .pickle extension. Which will be
used later in live feed facial recognition.
In the next step, the program opens the camera and start detection faces. When a face is detected,
the recognition process starts.
In one hand, if the detected face is being recognized. the name will be added to an attendance list in
addition to the detection time. and a graphical interface will appear showing personal information of
the detected person, those information is being extracted from an informations database designed
using sqlite. After 5 seconds the interface will vanish, and the program will return to detect new
faces.
In the other hand, if the detected face isn’t recognize. The program ignores that and returns to detect
new faces. However, if you not a student and want to attend the class anyways. You could manually
enter your informations. And these last ones will be added to the attendance list.
After that the program will return to the detection stage.

Ladies and gentlemen. I wanna thank you for your attention and leave you with a real life
demonstration.

You might also like