Skripsi

OPTIMASI HAND LANDMARK PADA HAND TRACKING MENGGUNAKAN CONVOLUTIONAL NEURAL NETWORK SEBAGAI AUGMENTED REALITY GAME CONTROLL

---

Penilaian

---

This study focuses on creating a flexible and lightweight solution of hand landmark prediction using Convolutional Neural Network. Beyond replicationg high-end VR/AR headsets’ hand tracking capabilities, the motivation extends to promoting gamer’s physical activity using augmented reality experiences. By using the hand landmark prediction as game controller for in-game actions, that encourages gamers to move their bodies, fostering a healthier and more active gaming experience. To achieve that, this study explores 3 configuration set of keypoint 11, 17, and 21 for maximizing the accuracy and speed in hand tracking. In addition, the methodology involves building an affordable and lightweight architecture based on U-Net. U-Net structure adapted to use Hourglass Network Block with depthwise convolutional layers and embedded pre and post processing layer. From the experiment, the model got 0.061 Mean Per Joint Position Error@128px and 14 to 16 dynamic frame rate score in 21 keypoint with low hardware i3 gen 7 CPU paired with MX130 GPU laptop. Keywords : Keypoint Prediction, Augmented Reality, Convolutional Neural Network (CNN), U-Net

---

Availability

Inventory Code	Barcode	Call Number	Location	Status
2407000663	T139024	T1390242023	Central Library (Referens)	Available but not for loan - Not for Loan

Detail Information

Series Title

-

Call Number

T1390242023

Publisher

Inderalaya : Prodi Teknik Informatika, Fakultas Ilmu Komputer Universitas Sriwijaya., 2023

Collation

xvi, 45 hlm.; Ilus.; 29 cm

Language

Indonesia

ISBN/ISSN

-

Classification

006.754 07

Content Type

Text

Media Type

unmediated

Carrier Type

-

Edition

-

Subject(s)

Prodi Teknik Informatika
Situs Jejaring Sosial

Specific Detail Info

-

Statement of Responsibility

SEPTA

Other version/related

File Attachment

Comments

---

You must be logged in to post a comment