Parsing Mediapipe Data for Face Mesh, Pose, and Hand Landmarks

YouTube video player

We all love the incredibly powerful features of the Mediapipe AI library. But, lets face it, the data is very hard to interpret, and the returned data structures are complex, and poorly documented. The purpose of this lesson is to show you how to parse the complex data structures coming from Mediapipe, and to create simple, intuitive arrays for the landmark data. We create classes in python which do the parsing, and the classes are easy to work with. With these classes, you can use mediapipe simply, and get simple to understand data structures back. The video explains what we are doing and how we are doing it. For your convenience the resulting code is posted below.

 

Parsing Mediapipe Data for Pose Landmarks, Hand Landmarks and Face Bounding Box


 

In this lesson we show how to create python classes to parse the data coming from Mediapipe for hand Landmarks, Pose Landmarks and the bounding boxes for found faces. Creating these classes allows the difficult parsing to be done in the class, and then have a simple way to parse and use all the data.

 

Python 3D Graphics Tutorial 21: Understanding and Using Widgets in Vpython

In this video Lesson, we show how to interact with a Python model using widgets. We show how Vpython allows use of dropdown menus, radio buttons, check boxes and slider bars. These combine to give the user great control over an animated model. In past lessons we changed parameters by editing the program. By incorporating widgets we can allow non-programmers to effectively interact with a model.

 

AI For Everyone: Parsing Pose and Hand Data Landmarks from Mediapipe

 In this video lesson we show you how to create simple Python classes to parse the data from mediapipe hand Landmarks and pose estimation. With these two classes, it is very easy to parse the most important data generated by mediapipe.

 

Distinguish Between Right and Left Hands in MediaPipe

 

In this video lesson we explore how to parse the data set returned from mediapipe to understand whether a given hand is a right hand or a left hand. Mediapipe has a method which determines the handedness of a found hand. We will show how to alter our data parsing class from earlier lessons to include the handedness of the found  hands. Enjoy!

 

Making The World a Better Place One High Tech Project at a Time. Enjoy!

Exit mobile version