Character Recognition through Gestures Using SVM Algorithm

摘要

The main objective of the paper is to develop a Machine Learning based system that uses gestures to perform functions. There are a lot of gesture based applications existing in today's world but every time we want to use a gesture controlled application we need to learn the predefined gestures and the functionality is also limited to the defaults provided with it. For this project, we aim to build a system which can be trained to recognize the gestures we make and perform the dedicated function we decide for it. We'll be demonstrating this by training the system to recognize letters by the gestures we make in air. To build this system, we'll be using an Arduino Board interfaced with an accelerometer. The device can be attached to the user's hand. The accelerometer will provide input to the microcontroller about the hand's (which is being used to make the gesture) coordinates. The algorithm will pick up this data and maintain a database to recognize each gesture differently. Once we train the system with the same gesture multiple times it will gather enough data to have an estimate of what the gesture should look like. This gesture can then be assigned to perform a task on the computer. We are also aiming to add Bluetooth support so as to make the gesture device wireless and more flexible.