Gesture recognition in mobile applications

399 views Asked by At

I am currently doing a project on mobile Android devices where I'd like to detect hand gestures with only RGB data on mobile devices. Those devices operate in various differing environments and may be moving. Therefore I cannot simply do a background subtraction. I tried a lot about skin color extraction in YCbCr and HSV (for convex hull analysis) but unfortunately I cannot prevent the background from having similar tones like skin. Do you have any suggestions on how to improve the segmentation or what different approaches might work out?

1

There are 1 answers

0
Fixus On

First of all you need to read about new google project, they are creating new smartphone with capera that will be able to recognize gestures (there will be depth camera and soft for that :) )

Using normal smartphone will be hear but I saw some simple samples that used OpenCV library for recognition movement and objects, Maybe this will be a hint for you