Air-swipe gesture recognition using OpenCV in Android devices

There is a need to enhance communication between human and computers which is being greatly defined in this changing era of technology with Human Computer Interaction (HCI) which is helping in determining new communication models and accordingly new ways of interacting with machines. Current smartphone inputs are limited to physical buttons, touchscreen input, cameras or built-in sensors. The rapid development of Smartphone's in the last decade was mainly due to interaction and visual innovations. For Example the given approaches of input either require a dedicated surface or Line-of-Sight for interaction. But in today's scenario of increasing computability of smartphone or other gadgets and their decreasing sizes have raised a need for such touch free operations over these gadgets. In such an intendment we introduce Air-Swipe Gesture Recognition System which can be useful to enable user to make In-Air gestures in front of the camera and to do different operations. This System can give a user-friendly and a live-experience of interaction and visualization, enhancing the usability and making the android device more interactive. It does not require any hardware changes instead only uses the native camera of the device and a machine learning software such as Open Source Computer Vision (OpenCV) algorithms to detect the changes in environment and respond accordingly in varying conditions. We tested this classification and found out the result that considering the frames to be divided into quadrants along x-axis and y-axis and found that the value of the frame matrix changes. Our approach has the capability of recognizing gestures with precision of almost 96%.