You Look Close to ?

• 

  demo

Inspiration

I was highly inspired by the way face recognizer in my phone works so thought to understand basic principles of face detection and recognition. It was a great learning process I understood several aspects of image processing and noise cleaning. After learning this I had also learned about neural networks and deep learning to train my model based on a set of data files.

What it does

The program detects the face from web cam and tries to recognize similar facial features from trained model. Giving you the best prediction about your closest resemblance.

How I built it

I basically used open cv haar_cascades to aid in face detection and determining "region of interest" (face). After identifying and labeling the faces from my raw data I generated a yml file having processed information. The face detector code than tries to match similarity with opencv LBPH face recognizer.

Challenges we ran into

I wasn't able to identify many of the regions of interest in my raw data so I had to change number of minimum neighbors to a lower value. Than second challenge I had to face was after installing opencv-python library there were several missing definitions for which I had to Install opencv-contrib library. To declare my face_recognizer.

Accomplishments that we're proud of

I was finally able to make a trained data file for my face detector.

What we learned

How to train data for facial recognition and how to process images for efficient and smooth recogniton.

What's next for You Look Close to ?

To extend the volume of data and develop an algorithm based on predicting future looks based on past time growth of input data(faces).

Built With

• numpy
• opencv
• python