CONVOLUTIONAL NEURAL NETWORK ARCHITECTURE OPTIMIZATION USING SQUEEZE AND EXCITATION ATTENTION FOR HERBAL LEAF CLASSIFICATION

UTOMO, RAGIL GIGIH (2026) CONVOLUTIONAL NEURAL NETWORK ARCHITECTURE OPTIMIZATION USING SQUEEZE AND EXCITATION ATTENTION FOR HERBAL LEAF CLASSIFICATION. Tugas Akhir thesis, University of Technology Yogyakarta.

Text 5220411255_Ragil Gigih Utomo_Abstrak.pdf Download (174kB)

Abstract

ABSTRACT The identification and classification of herbal leaves are crucial in various fields, including pharmaceuticals, botany, and traditional medicine. However, manual classification is often time-consuming and prone to errors. This study aims to develop an automated herbal leaf classification system by combining a Convolutional Neural Network (CNN) with Squeeze-and-Excitation Attention. The primary challenge addressed is the high variability in leaf shape and color. To overcome this, a CNN architecture enhanced with Squeeze-and-Excitation Attention is proposed to improve the model’s ability to extract relevant features from leaf images and focus on important regions. The research methodology involved collecting datasets of herbal leaf images, developing the model architecture, evaluating its performance, and building the system. The process included data augmentation, training a baseline CNN, integrating an attention mechanism, and optimizing hyperparameters. Test results showed that the proposed model achieved 86% classification accuracy on the herbal leaf dataset. In summary, this study successfully demonstrates that integrating the attention mechanism into a CNN significantly improves the classification performance of herbal leaf images, offering substantial potential for practical applications in species identification and quality control. Keywords: Convolutional Neural Network, Herbal Leaves, Deep Learning, Squeeze and Excitation Attention

Item Type:	Thesis (Skripsi, Tugas Akhir or Kerja Praktek) (Tugas Akhir)
Subjects:	T Technology > T Technology (General) T Technology > T Technology (General) > T201 Patents. Trademarks
Divisions:	Fakultas Sains Dan Teknologi > S1 Informatika
Depositing User:	Kaprodi S1 Informatika UTY
Date Deposited:	07 May 2026 03:30
Last Modified:	07 May 2026 03:30
URI:	http://eprints.uty.ac.id/id/eprint/19828

Actions (login required)

View Item