Face-Recognition

An attempt at building a face recognition program with the help of OpenCV.

To be honest the heavy lifting has been done by OpenCV itself and in this project I have only utilised the tools already available to build something useful

3 STEP PROCESS

CREATE DATA
TRAIN MODEL
RECOGNITION

STEP 1 : CREATING THE DATASET

First of all we need to create labelled dataset so that the model can refer to it for labelling new faces it encounters; this is handled by the create_data script

just run the script and user will be prompted to provide a name (a unique label to classify the observed face as!)
upon entry of a unique name for the face a preview screen will pop up display the camera view with the face enclosed in a box.
at this poiint the script is recording images of the face detected, stop after sometime (20 seconds of waiting is more than enough)
in the end simply close the script.

At the end of this step you will find that images of your face have been captured and saved in the ./dataset/ directory

STEP 2 : TRAINING MODEL AND GENERATING THE FEATURE SET

Now that we have sufficient data about ( prefferably multiple ) individual faces we go ahead to train our model so that the distinguishing features of the faces can be extracted.

in this step we simply run the train_model.py script, which in the end generates a pickle dump (consider pickle as a format to store data)
simply run the script and you will find the images being analysed and data stored in ./encodings/encoding.pickle (all the faces are stored in the same pickle file)

STEP 3 : TESTING YOUR MODEL

Since the model has trained itself and now knows about the facial data of some individuals. You can try to test it's capability by running the recognition.py script

the user will get a preview of the camera view with the faces detected being in a box with on top

resource utilised:

-> OpenCV ( https://github.com/opencv/opencv )
-> Face_Recognition ( https://github.com/ageitgey/face_recognition )

nikhil-k-singh / face-recognition Goto Github PK