Skripsi

PENGENALAN BAHASA ISYARAT INDONESIA SECARA REALTIME MENGGUNAKAN RECURRENT NEURAL NETWORK

---

Penilaian

---

Hand gestures are one of the communication media that people with disabilities can use to communicate. In daily life, deaf people often have difficulty communicating with normal people because not everyone understands sign language. Thus, a sign language interpreter that allows people who do not know sign language to communicate with deaf people is needed. The methods that are often used today are still limited to the Indonesian Sign Language System (SIBI), while the common language used to communicate is Indonesian Sign Language (BISINDO). In addition, the method used is also still very dependent on the accuracy of feature extraction. So, in this research a system was developed that is used to recognize BISINDO using the Recurrent Neural Network (RNN) method. The data used in this study was taken with a webcam in the form of a video that will be converted into frames and arrays. Data was taken from 3 respondents with a total of 3.240 videos and 97.200 array data consisting of letter and number characters. From the parameters that have been tested, the training results show that the use of the Adam optimizer with a learning rate of 0.0001, and 500 epochs shows the best accuracy with a minimum loss value compared to testing using other parameters. This model was then used in realtime testing which was carried out 5 times for 36 classes. The accuracy result obtained is 81.67%. The error that occurs can be caused by the similarity of the existing hand signal language, such as the letters I, J, D, P, M, and N

---

Availability

Inventory Code	Barcode	Call Number	Location	Status
2207002706	T75664	T756642022	Central Library (Referens)	Available

Detail Information

Series Title

-

Call Number

T756642022

Publisher

Indralaya : Prodi Teknik Elektro, Fakultas Teknik Universitas Sriwijaya., 2022

Collation

xvi, 43 hlm.; ilus.; 29 cm

Language

Indonesia

ISBN/ISSN

-

Classification

621.382 07

Content Type

Text

Media Type

-

Carrier Type

-

Edition

-

Subject(s)

Teknik elektro
communication network analysis

Specific Detail Info

-

Statement of Responsibility

MURZ

Other version/related

File Attachment

Comments

---

You must be logged in to post a comment