Creating your own Snapchat filter using OpenCV
Sarvesh Shroff (~sarvesh77) |
Have you ever wondered how snapchat filters work? In this talk we will give you a thorough explanation and demo of the famous face swap filter using OpenCV, dlib and NumPy.
We will do a line-by-line walkthrough of our code to extract facial landmarks of both images using methods like convex hull and delaunay triangulation. We then swap faces of the two input face images and blend them using the seamlessclone method. We will also go through various computer vision concepts required to understand the underlying mathematics.
After this talk you would be able to learn how to do the above mentioned tasks and some insights into a few OpenCV methods and we will also go over a little bit of numpy basics.
Introduction and live demo [5 min]
Explanation of facial landmark detection methods [5 min]
Overview of functions used in our code [5 min]
Line by line walkthrough of the code [10 min]
Questions from the audience [5 min]
Love for Python, Familiarity with Python3 syntax
Example of one of our outputs:
I am currently a sophomore at SRM University AP and a Researcher at Next Tech Lab, A QS Reimagine Award-winning student-run innovation lab. I have won a national level robotics championship held at IIT-R.
I am MTH Junaidi, sophomore at SRM University AP and a Researcher at Next Tech Lab, A QS Reimagine Award-winning student-run innovation lab. Also gave a lightning talk at PyCon Taiwan 2018