OpenCV for Android accelerometer/gyroscope gesture recognition?

asked 2014-11-20 01:49:25 -0600

xetiro gravatar image

updated 2014-11-20 06:11:39 -0600


I am writing an application (for academic/research purpose) for Android smartphones that needs to recognize a small set of motion gestures performed by the user (triangle, circle, square, cross, etc) .

Motion gestures typically use accelerometer and gyroscope data.

I already wrote an app that will collect data samples provided by users. This app will ask users to perform a small set of motion gestures with their Android phones while the app reads and collects accelerometer and gyroscope data from the phone. This data will then be used to train a classifier in order to use in the final app that should be able to recognize the gestures made by the users.

So I need some machine learning framework to implement the recognition facility on my app. And I am wondering if OpenCV is suitable for this.

Could you clarify me if OpenCV is the right tool for such a system?

edit retag flag offensive close merge delete


It could work with OpenCV. OpenCV has a machine learning toolbox that would be suitable for the learning part, however not all recent learning techniques are included. Can you specify in what direction you are looking? Boosting, SVM, forests, ... ?

StevenPuttemans gravatar imageStevenPuttemans ( 2014-11-20 03:52:43 -0600 )edit

I have been reading some papers about motion gesture (i.e. accelerometer) recognition. It seems that the most common approach is to use Hidden Markov Models. Do OpenCV support this?

xetiro gravatar imagexetiro ( 2014-11-20 23:53:40 -0600 )edit