Facial Recognition-based Security Authorization

Inspiration

We recognize that facial-recognition is at the _ forefront _ of cyber-security technology and that by creating a program to link basic machine authorization to trained facial features, we can stop millions of cyber attacks every day.

What it does

This project uses supervised machine-learning to train a classifier to recognize and distinguish one face from the rest. Further, our algorithm creates a LBPHFaceRecognizer, or Localized Binary Patterns Histogram recognizer, that makes a model for new test data. Then, we use live feed from a web cam to either authorize or reject a person from the system. If the algorithm rejects the person, we capture the image for the proper owner's security and use image processing to make descriptions of the person (their smile and width of a face).

How we built it

We built the Security Authorization project using both OpenCV and dlib libraries. These provided us with face-recognition cascade classifiers, with large databases of faces. Additionally, they provided functionality to the creation of the model and pick out key features of the face.

Challenges we ran into

We ran into issues in setting up dlib and trying to use the HaarCascades to recognize a face in an image.

Accomplishments that we're proud of

The algorithm is highly accurate in using the trained data to perform facial recognition and is accurate in making descriptions of the users who were not being trained for.

What we learned

We learned how to use OpenCV and build code that follows the structure of a machine learning algorithm.

What's next for Facial Recognition-based Security Authorization

For future projects, we plan on linking the correct facial recognition to an actual hardware authorization and being able to make more descriptions of a face with dlib!

Built With

• dlib
• opencv