version 1.1
This application gets the geographic coordinates from the five fingers of the right hand and saves these points on a .txt file. Later on, there is a parser to format the file and allow it to be used for training and classification.
We've used it to capture some gestures of Brazilian Sign Language.
The Kinect-Finger-Tracking API, developed by LightBuzz Software tracks the extremity of 5 fingers of each hand. All the data is get from Kinect V2 using the principles of Convex Hull.
Inicially, we've used the hand's center point - information provided by Kinect:
We need to use the center point to get the medium point - the intersection between intermediate phalanges and proximal phalanges:
Once the area to be used on calculus is defined, the extraction of these points is through their angulation. This is used to fix any possible noise generated by the angle of the middle finger.
On the end we have the following points:
- Inessa Luerce - @inessadl
- Lucas Tortelli - @lmtortelli
- Carpals, Metacarpals, and Phalanges (The hand)
- Coordinate mapping
- LightBuzz/Kinect-Finger-Tracking
- LightBuzz/Vitruvius
- Scikit-Learn: Machine Learning in Python
Licensed under the Apache License, Version 2.0