SlobodaStudio-face-recognition

Solution algorithm:

1. Initially we are preparing a photos database of people whom we will recognize in the future
2. When the database is ready, we connect the video stream from the webcam and use cnn to detect the faces in the photo
3. Then, with the help of another cnn, we highlight landmarks on the face
4. Then we compare them with already known templates

Technologies that have been applied

For our project we don't need powerful and expensive hardware, only compact raspberry pi and camera.

1. cvlib
   
2. dlib
   
3. face-recognition
   
4. Keras
   
5. matplotlib
   
6. numpy
   
7. opencv-contrib-python
   
8. Pillow
   
9. scikit-learn
   
10. tensorboard
    
11. tensorflow
    

Demonstration of the algorithm, click on the image to go to the video viewing

In this video you can watch the work of the CNN that created and trained by us to recognise the actors of the "Justice League"