Machine Learning project to recognise people from an Image just like facebook.
Built with the help of dlib's state-of-the-art face recognition built with deep learning. The model has an accuracy of 99.38% on the Labeled Faces in the Wild benchmark.
-
Python 3.x
-
Numpy
-
Scipy
-
Tip: Installing dlib can be a tedious job. On macOS or Linux you may follow this link.
-
Extras:
-
OpenCV (required only in
webcam.py
for capturing frames from the webcam) -
For using
./demo-python-files/projecting_faces.py
you will need to install Openface.To install Openface, follow the below instructions:
$ git clone https://github.com/cmusatyalab/openface.git $ cd openface $ pip install -r requirements.txt $ sudo python setup.py install
-
- Clone this repository
git clone [email protected]:anubhavshrimal/FaceRecognition.git
.
-
Make folder
training-images
. -
Add images of each person you want to recognise to a folder by their name in
training-images
.Example
$ mkdir training-images $ cd training-images $ mkdir Name_Of_Person
Then copy all the images of that person in
./training-images/Name_Of_Person
folder. -
Run on cmd
python create_encodings.py
to get the encodings of the images and the labels. This will createencoded-images-data.csv
andlabels.pkl
files.Note: There has to be only one face per image otherwise encoding will be for the first face found in the image.
-
Run on cmd
python train.py
to train and save the face recognition classifier. This will createclassifier.pkl
file. It will also createclassifier.pkl.bak
backup file if the classifier with that name already exists.
-
Make folder
test-images
which contains all the images you want to find people in. -
Run on cmd
python predict.py
to predict the faces in each image.
- Thanks to Adam Geitgey whose blog inspired me to make this project.
- Thanks to Davis King for creating dlib and for providing the trained facial feature detection and face encoding models used in this project.