Just a couple of tests for trying to make a facial recognition program that correctly classifies different people's faces
Before you download, be advised! This repo contains the training image dataset, which is composed of more than 50 MBs of face pictures.
It currently looks like this:
And its performance in terms of predictions' confidence is this:
(as you can see, it gets a little bit confused :^])
Before anything can run, you need to create a Python virtual environment & install the dependencies:
$ python -m venv env
$ pip install -r requirements.txt
Note: for installing the dlib library, follow this tutorial or this article.
Firstly, run the training program within this repo, locating all faces inside the faces/ directory, and naming each person's faces folder accordingly (just immitate the given folder structure):
$ python faces-train.py
Then, go ahead and run the webcam-capture recognizer:
$ python cv.py
You can get statistic graphs generated after you quit the program if you run it with the -p/--plot flag:
$ python cv.py -p
(a full list of options can be obtained with the -h/--help flag)