PENGENALAN EKSPRESI WAJAH MENGGUNAKAN KOMBINASI EKSTRAKSI FITUR LBP DAN 3 TIPE KONEKSI MULTI-LEVEL CNN

Septyono, Muhammad Bagas (2025) PENGENALAN EKSPRESI WAJAH MENGGUNAKAN KOMBINASI EKSTRAKSI FITUR LBP DAN 3 TIPE KONEKSI MULTI-LEVEL CNN. Undergraduate thesis, Universitas Pembangunan Nasional "Veteran" Jawa Timur.

	Text (Cover) 21081010049_Cover.pdf Download (1MB)
	Text (Bab 1) 21081010049_Bab 1.pdf Download (284kB)
	Text (Bab 2) 21081010049_Bab 2.pdf Restricted to Registered users only until 18 June 2028. Download (868kB) \| Request a copy
	Text (Bab 3) 21081010049_Bab 3.pdf Restricted to Registered users only until 18 June 2028. Download (776kB) \| Request a copy
	Text (Bab 4) 21081010049_Bab 4.pdf Restricted to Registered users only until 18 June 2028. Download (2MB) \| Request a copy
	Text (Bab 5) 21081010049_Bab 5.pdf Download (249kB)
	Text (Daftar Pustaka) 21081010049_Daftar Pustaka.pdf Download (191kB)

Abstract

Facial expressions are essential non-verbal cues that reflect a person's emotional and psychological state. However, automatic facial expression recognition still faces various challenges, including uneven lighting, varying head poses, and background noise. This study aims to enhance facial expression classification accuracy by combining a Multi-Level Convolutional Neural Network (MLCNN) with Local Binary Pattern (LBP). The MLCNN is designed to extract hierarchical features—low-level, mid-level, and high-level—while LBP captures important local texture information. The model is evaluated using the FER2013 dataset, which contains seven categories of facial expressions: angry, happy, sad, surprise, disgust, fear, and neutral. Eight experimental scenarios were conducted, involving variations in the connection types between MLCNN blocks and the integration of LBP as an additional input branch. The best result was achieved using the combined LBP + MLCNN model with connection type 3, yielding an accuracy of 68.7% and an F1-score of 67.1%, outperforming the conventional CNN model. Moreover, the LBP + CNN configuration also showed improved performance, indicating that local texture features contribute significantly to expression classification. This research demonstrates that integrating LBP and MLCNN enhances model robustness against real-world variability in facial expressions. The model was also successfully implemented in a real-time expression detection system using a camera, highlighting its potential applications in human-computer interaction and emotional condition monitoring.

Item Type:

Thesis (Undergraduate)

Contributors:

Contribution	Contributors	NIDN/NIDK	Email
Thesis advisor	Anggraeny, Fetty Tri	NIDN0711028201	fettyangraeny.if@upnjatim.ac.id
Thesis advisor	Mumpuni, Retno	NIDN0016078703	retnomumpuni.if@upnjatim.ac.id

Subjects:

Q Science > QA Mathematics > QA75 Electronic computers. Computer science

Divisions:

Faculty of Computer Science > Departemen of Informatics

Depositing User:

Muhammad Bagas Septyono

Date Deposited:

19 Jun 2025 01:20

Last Modified:

19 Jun 2025 01:20

URI:

https://repository.upnjatim.ac.id/id/eprint/38512