This system uses Python 3 and OpenCV to recognize faces in Raspberry-Pi's video-streams sent via tcp protocol, while using the ImageZMQ classes (https://github.com/jeffbass/imagezmq), to a central computer to process all the data. The system also save frames of unknown faces and use them to train a new dataset of known faces to track every individual on the environment.