Face Applications

This project contains face related applications using opencv and deep learning libraries.

Face Recognition using LBPH Face Recognizer

The pipeline involves the following steps :-

• Detecting faces using OpenCV's Haar Cascades
• Training a Face Recognizer using OpenCV's build-in LBPH Face Recognizer
• Saving the model and predicting on custom images

Click here to access the code.

Face Recognition using One Shot Learning

On a contrary to the above Face Recognition method which involves training on several images, One Shot learning only requires a pre-trained to directly get the embeddings from the detected face. The code involves the following steps :-

• Detecting faces using MTCNN Face Detector
• Getting feature embeddings of each class from pre-trained VGGFace model and storing them in a database
• Get the feature embeddings of the test image and calculate the distance between this test image and each embedding in the database
• If the distance is less than a threshold for a particular pair, then the identity has been found

Click here to access the code

Face Alignment

It is one of the most important steps in modern day Face Recognition pipeline to increase the accuracy of recognition.

Click here to access the code.