General Remote Control Based on Hand Patterns Detection Using Convolutional Neural Network

Abstract

In this paper, the system is made to be able to recognize hand patterns to be able to control electronic devices, as a substitute for the conventional remote. This system consists of a computer containing a program with a trained Convolutional Neural Network (CNN) model and an MCU node containing a remote infrared signal database whose function we want to replace. The camera functions as a reader of hand patterns which will be classified using trained CNN models. Each hand gesture represents the infrared pulse in the database program n Node MCU. There are 4 hand patterns that represent commands in this proposed system. To be able to handle much more commands, this proposed system is designed to change mode according to the combination of hand patterns detected by the system algorithm. The changing of mode results in the changing of the infrared pulse which is represented by each hand pattern in the database. Node MCU will transmit infrared pulse corresponds to hand gesture from CNN classification after receiving a command from the computer through TCP/IP communication.

Rancangan Bangun Sistem Pembatas Jumlah Penumpang Pada Lift Dengan Metode Klasifikasi K-Nearest Neighbors (Knn)

“…Pada tahap ini citra yang sebelumnya sudah dirubah kedalam bentuk citra grayscale dengan proses background substraction di ubah kembali ke dalam bentuk biner [11]. Citra biner merupakan citra yang membutuhkan 1 bit data yaitu 0 dan 1 dari citra biner.…”

“…Opening dan closing merupakan teknik yang dapat minimalisir noise setelah dilakukannya proses background substraction. Opening fungsinya untuk mengikis atau mengeliminasi piksel yang tidak dibutuhkan [11]. Closing fungsinya untuk mengisi lubang pada piksel.…”