While we enjoy snapchat filters in our hours of boredom, the technology behind it is much more advanced and complex. How does the app instantly recognize the face and rightly puts the filter?

The application is a product of Computer Vision technology that works with the facial recognition software. Started by a Ukrainian startup, the app allowed the user to change their facial features for images or video chat. After acquiring the startup, Snapchat started working on we all now use in our daily life. Snapchat filters use Image processing as a specific area for CV. It transforms the image by performing an algorithm on every pixel of the picture. The app follows and works on the following elements:

Face Detection

While face detection has been in the scene since the 2000s, detection of very tiny pixels from the front angle still posed some challenges. Support Vector Machine and Histogram Oriented Gradients was the most widely used technology for this, providing good quality from real-time detection.

Facial Landmark

Next comes the analysis work. For every detected face, there is coordination between the local region for every member or feature including mouth, nose, lips, bone and eyes. Detection and landmark extraction is combined in one operation.

Image Processing

Snapchat filters use Image processing as a specific area for CV. It transforms the image by performing an algorithm on every pixel of the picture.

Filter to Face swap

When the application locates all the features of the two people, it creates a mesh which sticks frame by frame with every point, leading to modification by snapchat; in this case, a face swap.

This technology that Snapchat is working on is still relatively new, but more work has been done to perfect it over the years. The achievement itself is that it is accessible to every regular person and we now understand why Snapchat paid more than a million dollars to acquire it.