/ Machine Learning

A Face Recognition system that works in real time

A Face Recognition system that works in real time

Face-Recognizer

A Face Recogniser which works in real time!
You get 3 options:

1) Add a new face to the dataset of known people :heavy_exclamation_mark:.

Important: Make a folder with the name 'people_folder' in the directory of this code and then proceed.

Enter the name of the person and press ENTER when ready.

A demo of this is shown below :wink::

add_face

Photos of the person are clicked and then saved in a folder with the person's name as the folder name.

added_faces

2) This option is for live recognition. :smiley:

Our recognition system works!!

added_faces

3) Finally on pressing 3, the exe file stops and you exit! :v:


How does it work :question:

1) It takes in 20 images per face(person). :camera:

i) Finds the face in the frame using a HAAR cascade.

ii) Trims the unnecessary parts of the face.

iii) Does histogram equalization and resizes the images to 100 x 100.

These images are saved in a folder with the person's name.

2) For live recognition: :+1:

i) Creates a LBHP face recogniser and trains it on the existing dataset.
ii) Finds faces in live video stream, does the same pre-processing as point 1.
iii) Finds the face in our dataset with the closest likeness to the current face within a certain threshold.
iv) Displays the face along with the name of the person and draws a rectangle around their face.

GitHub