Membangun Virtual Pen Berbasis Hand Pose Menggunakan Convolutional Neural Network

Riskiana, Meril Lia Priday (2023) Membangun Virtual Pen Berbasis Hand Pose Menggunakan Convolutional Neural Network. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 07211940000022-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 September 2025. Download (10MB) | Request a copy

Abstract

Seiring berkembangnya teknologi, kegiatan menulis yang menggunakan papan tulis pada saat kegiatan belajar mengajar sekarang ini dilakukan melalui monitor komputer atau layar laptop. Teknologi yang umum digunakan untuk menginputkan data text berupa tulisan tangan pada komputer yaitu stylush pen dan pen tablet. Stylush pen hanya dapat berfungsi pada layar yang mendukung teknologi touchscreen. Luas tab pada pen tablet membatasi keleluasaan pengguna dalam menulis. Selain itu, fokus pengguna terbagi karena melihat layar laptop atau monitor komputer dan tab secara bergantian. Virtual pen merupakan suatu sistem yang mampu menginputkan tulisan tangan berdasarkan hand pose. Terdapat penelitian pengembangan virtual paint dengan fungsi yang sama, akan tetapi sistem klasifikasi hand pose masih dilakukan secara manual dengan pendekatan perhitungan jumlah jari tangan yang sedang berdiri. Pada penelitian ini akan dilakukan perbaikan dengan menggunakan algoritma convolutional neural network untuk melakukan klasifikasi terhadap hand pose. Tahapan dilakukan dari input berupa citra gambar hand pose menggunakan kamera berlanjut ke tahap klasifikasi hand pose kemudian adalah proses voting pengendali fungsi pada virtual pen dan yang terakhir adalah proses pengendalian fungsi menggunakan hand pose. Berdasarkan tahapan yang telah dilaksanakan, dihasilkan akurasi klasifikasi dengan nilai optimum pada jarak 40 cm dari kamera yaitu sebesar 91.2%. Nilai akurasi tersebut diperoleh dari pengujian yang dilakukan terhadap lima orang responden.
====================================================================================================================================
As technology develops, writing activities that use blackboard on when teaching and learning activities are now carried out through a computer monitor or laptop screen. The technology commonly used to input text data is in the form of handwriting on computers, namely stylus pen and tablet pen. Stylus pen can only function on a screen that supports touchscreen technology. The tab area on the pen tablet limits flexibility on writing. Besides that, the user’s focus is divided by looking at the screen and tab. Virtual pen is a system to input handwriting based on hand pose. There is research on the development of virtual paint with the same function, but the hand pose classification system is still being implemented manually with the approach of calculating the number of fingers that are standing. On This research will be improved by using a convolutional neural network algorithm to classify hand pose. The stages are carried out from the input in the form of images of hand pose using the camera continuing to the hand pose classification stage, then is the process of voting for the function control on the virtual pen and the last is the process functional control using hand pose. Based on the stages that have been carried out, it is generated classification accuracy with the optimum value at a distance of 40 cm from the camera is 91.2%. Accuracy value was obtained from tests conducted on five respondents.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Convolutional Neural Network, Hand Pose, Mediapipe, Virtual Pen
Subjects:	Q Science > QA Mathematics > QA336 Artificial Intelligence Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science) T Technology > T Technology (General) > T57.5 Data Processing
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Computer Engineering > 90243-(S1) Undergraduate Thesis
Depositing User:	Meril Lia Priday Riskiana
Date Deposited:	01 Sep 2023 04:10
Last Modified:	01 Sep 2023 04:10
URI:	http://repository.its.ac.id/id/eprint/101837

Actions (login required)

View Item