Prediksi Huruf Semaphore Berbasis Pose Menggunakan Deep Learning

Wibisono, Muhammad Daffa Zahrandika (2023) Prediksi Huruf Semaphore Berbasis Pose Menggunakan Deep Learning. Other thesis, Institut Teknologi Sepuluh Nopember.

[thumbnail of 07211940000064-Undergraduate_Thesis.pdf] Text
07211940000064-Undergraduate_Thesis.pdf - Accepted Version
Restricted to Repository staff only until 1 October 2025.

Download (9MB) | Request a copy

Abstract

Sistem komunikasi visual menggunakan semaphore telah lama digunakan dalam berbagai bidang, seperti militer, kapal laut, dan sektor transportasi. Metode ini melibatkan penggunaan bendera atau tangan mengirim pesan antara dua atau lebih pihak. Namun, penggunaan semaphore secara tradisional membutuhkan latihan dan keahlian khusus, serta kemampuan memahami kode semaphore yang kompleks.Penelitian ini bertujuan mengembangkan sistem rekognisi pose semaphore menggunakan MediaPipe dan klasifikasi Deep Learning melalui metode Convolutional Neural Network (CNN). Proses pengembangan sistem melibatkan langkah-langkah seperti pengumpulan dan pengolahan dataset, ekstraksi pose menggunakan MediaPipe, dan klasifikasi pose menggunakan model-model CNN seperti CNN, CNN ResNet50V2, dan CNN Xception.Hasil penelitian ini menunjukkan bahwa sistem yang dikembangkan mampu mendeteksi huruf dan kata dalam pose semaphore dengan tingkat akurasi yang tinggi. Model CNN ResNet50V2 mencapai akurasi 88.36%, Model CNN Xception mencapai akurasi 88.36%, Model CNN mencapai akurasi 93.72%, dan Model CNN2 mencapai akurasi 93.53%.Berdasarkan temuan ini, sistem rekognisi pose semaphore memiliki potensi pemanfaatan dalam pendidikan di sekolah. Dengan bantuan teknologi ini, pembelajaran tentang semaphore dapat lebih menarik dan interaktif. Siswa dapat belajar mengenai
komunikasi visual melalui semaphore dengan bantuan sistem yang mampu mendeteksi dan mengenali pose huruf dan kata. Penggunaan sistem rekognisi pose semaphore dalam pendidikan juga dapat memberikan kemungkinan eksplorasi lebih lanjut dalam pengembangan metode pembelajaran interaktif lainnya

============================================================================================================================================

The visual communication system using semaphore has been widely used in various fields such as the military, maritime, and transportation sectors. This method involves the use of
flags or hand signals to transmit messages between two or more parties. However, traditional semaphore usage requires special training, expertise, and the ability to understand complex semaphore codes.This research aims to develop a semaphore pose recognition system using MediaPipe and Deep Learning classification through Convolutional Neural Network (CNN) methods. The development process involves steps such as dataset collection and processing, pose extraction using MediaPipe, and pose classification using CNN models such as CNN, CNN ResNet50V2, and CNN Xception.The results of this research demonstrate that the developed system is capable of accurately detecting letters and words in semaphore poses. The CNN ResNet50V2 model achieved an accuracy of 88.36%, CNN Xception achieved 88.36%, CNN achieved 93.72%, and CNN2 achieved 93.53%. Based on these findings, the semaphore pose
recognition system has potential applications in education, particularly in schools. With the help of this technology, learning about semaphore can become more engaging and interactive. Students can learn visual communication through semaphore with the assistance of a system that can detect and recognize letter and word poses. The use of the semaphore pose recognition system in education also opens up possibilities for further exploration in the development of
other interactive learning methods.

Item Type: Thesis (Other)
Uncontrolled Keywords: Semaphore, MediaPipe, Convolutional Neural Network , Pose Tubuh, Deep Learning , Xception , ResNet50V2 , Human Pose
Subjects: Q Science > QA Mathematics > QA76 Computer software
Q Science > QA Mathematics > QA76.754 Software architecture. Computer software
Divisions: Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Computer Engineering > 90243-(S1) Undergraduate Thesis
Depositing User: Muhammad Daffa ZW
Date Deposited: 18 Jan 2024 02:43
Last Modified: 03 May 2024 02:22
URI: http://repository.its.ac.id/id/eprint/102211

Actions (login required)

View Item View Item