Deteksi Orang Jatuh Berbasis 3D Convolutional Neural Network

Fauzan, Hafizh (2021) Deteksi Orang Jatuh Berbasis 3D Convolutional Neural Network. Undergraduate thesis, Institut Teknologi Sepuluh Nopember.

[img] Text
07211740000019-Undergraduate_Thesis.pdf - Accepted Version
Restricted to Repository staff only until 1 October 2023.

Download (18MB) | Request a copy

Abstract

Jatuh adalah penanda kelemahan, imobilitas, dan gangguan kesehatan akut dan kronis pada orang tua. Bahkan ketika cederanya tidak begitu serius, lansia seringkali kesulitan untuk bangkit tanpa bantuan ,terkadang mengarah ke 'long-lie' di mana lansia tetap terjebak di lantai untuk periode waktu yang lama. 'Long-lie' dapat menyebabkan dehidrasi, Ulkus dekubitus, pneumonia, hipotermia dan kematian. Pada Tugas Akhir ini akan dikembangkan sebuah sistem yang dapat mendeteksi jatuh pada manusia lanjut usia menggunakan algoritma 3D Convolutional Neural Network. Adapun training data yang digunakan berasal dari beberapa dataset publik dan akan dibuat dataset pribadi untuk testing dan revisi sistem. Sistem memiliki dua komponen utama, IP Camera (kamera yang dapat dihubungkan dalam suatu jaringan secara wireless) dan komputer untuk melakukan pemrosesan. IP Camera dan komputer terhubung pada suatu jaringan yang sama. Dari hasil pengujian didapatkan akurasi tinggi pada percobaan dengan jatuh ke depan sebesar 88.89%. Untuk jatuh ke belakang memiliki akurasi sebesar 55.56%. Sedangkan pada jatuh ke samping memiliki akurasi rendah sebesar 11.11% dan jatuh dari kursi memiliki akurasi sebesar 22.22%. ======================================================================================================= Falls are markers of weakness, immobility, and acute and chronic health problems in the elderly. Even when the injury is not serious, the elderly often struggle to get up unaided, sometimes leading to a 'long lie' in which the elderly remains trapped on the floor for long periods of time. 'Long-lie' can cause dehydration, decubitus ulcers, pneumonia, hypothermia and death. In this final project, a system that can detect falls in elderly people will be developed using the 3D Convolutional Neural Network algorithm. Meanwhile, the training data used comes from several public datasets and private datasets will be created for system testing and revision. The system has two main components, IP Cameras (cameras that can be connected in a network wirelessly) and computers to perform processing. IP Cameras and computers are connected to the same network. From the test results obtained high accuracy in the experiment with falling forward of 88.89%. To fall backwards has an accuracy of 55.56%. While falling to the side has a low accuracy of 11.11% and falling from a chair has an accuracy of 22.22%.

Item Type: Thesis (Undergraduate)
Uncontrolled Keywords: Jatuh, Deteksi, Sistem, Fall, Detection, System
Subjects: Q Science > Q Science (General) > Q325.5 Machine learning.
Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science)
Divisions: Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Computer Engineering > 90243-(S1) Undergraduate Thesis
Depositing User: Hafizh Fauzan
Date Deposited: 04 Sep 2021 07:53
Last Modified: 04 Sep 2021 07:53
URI: https://repository.its.ac.id/id/eprint/91408

Actions (login required)

View Item View Item