Melola Discovery

← Melola Discovery

Akses terbuka Jurnal

AI-ASSISTED WEB-BASED LEARNING MEDIA FOR HERBAL LEAF IDENTIFICATION: DEVELOPMENT, MODEL EVALUATION, AND USABILITY TESTING

Jurnal Ilmiah Ilmu Terapan Universitas Jambi · Vol. 10 No. 5 (2026) · hlm. 2524–2535

DOI
10.22437/jiituj.v10i5.58527
ISSN
2580-2259, 2580-2240
Terbit
2026-10-31
Penerbit
LPPM Universitas Jambi
Bahasa
en
Lisensi
creativecommons.org/licenses/by/4.0

Abstrak

The growing use of herbal plants highlights the need for effective learning media that help vocational students identify medicinal plant species more easily. This study aimed to develop an AI-supported, web-based learning tool for medicinal plant leaf identification, evaluate the performance of lightweight Convolutional Neural Network (CNN) models, and assess the tool's feasibility and usability. We used a Research and Development (R&D) approach based on the ADDIE model. We prepared a dataset of 3,100 images across 20 classes to compare four pre-trained lightweight CNN architectures: MobileNetV2, MobileNet, EfficientNet-B0, and EfficientNetV2-B0. The selected model was integrated into the web-based learning tool via a client-server architecture using an Application Programming Interface (API). Model performance was evaluated using accuracy, loss, precision, recall, and F1-score metrics, while feasibility was assessed by media and subject matter experts. 29 vocational students evaluated usability using the System Usability Scale (SUS). EfficientNetV2-B0 achieved the highest average accuracy (98.67%) across three random seeds, and the final model attained 97.25% accuracy on the test set. The tool received feasibility scores of 83.41% from media experts and 96.19% from subject matter experts. The average SUS score was 79.14, indicating good usability from the users' perspective. These findings demonstrate that integrating lightweight CNN models into web-based learning tools can provide a practical resource for learning to identify medicinal plant leaves. The developed tool can support students in both independent and classroom learning, although its broader implementation remains constrained by dataset diversity, device compatibility, internet dependency, and the limited scope of user evaluation.

Kata kunci

Jaringan kolaborasi para penulis

10 penulis · 16 tautan kepenulisan bersama · 9 lintas organisasi

Arahkan kursor atau ketuk sebuah nama untuk menyorot rekan penulisnya. Klik dua kali untuk membuka halaman penulis.

  • Universitas Pembangunan Nasional Veteran Yogyakarta
  • Institut Teknologi Adhi Tama Surabaya
  • Sebelas Maret University
  • Khon Kaen University
  • State University of Malang
  • Institut Teknologi Sepuluh Nopember
  • Organisasi berbeda
  • Organisasi sama
  • Afiliasi tidak diketahui
  • Garis lebih tebal = lebih banyak artikel bersama · lingkaran lebih besar = lebih banyak artikel
Tampilkan sebagai tabel
Rekan penulisOrganisasiArtikel bersama
Kartika Candra KiranaState University of Malang1
Tri Hadiah MuliawatiInstitut Teknologi Sepuluh Nopember1
Luky Agus HermantoInstitut Teknologi Adhi Tama Surabaya1
Vynska Amalia PermadiUniversitas Pembangunan Nasional Veteran Yogyakarta1
Dian Indri PurnamasariUniversitas Pembangunan Nasional Veteran Yogyakarta1
Riza Prapascatama AgusdinUniversitas Pembangunan Nasional Veteran Yogyakarta1

Afiliasi berasal dari OpenAlex; metadata OAI-PMH jurnal tidak memuatnya. Penulis tanpa ID OpenAlex dicocokkan berdasarkan nama, sehingga dua orang bisa tergabung atau satu orang bisa terpisah.

Sumber metadata: OAI-PMH (metadata jurnal) https://online-journal.unja.ac.id/index.php/JIITUJ/oai, pertama di-harvest 5 Nov 2026, terakhir terlihat 15 Nov 2026.
Metadata asli (oai_dc)
<oai_dc:dc xmlns="http://www.openarchives.org/OAI/2.0/" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:oai_dc="http://www.openarchives.org/OAI/2.0/oai_dc/" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://www.openarchives.org/OAI/2.0/oai_dc/  http://www.openarchives.org/OAI/2.0/oai_dc.xsd">
	<dc:title xml:lang="en">AI-ASSISTED WEB-BASED LEARNING MEDIA FOR HERBAL LEAF IDENTIFICATION: DEVELOPMENT, MODEL EVALUATION, AND USABILITY TESTING</dc:title>
	<dc:creator xml:lang="en">Liantoni, Febri</dc:creator>
	<dc:creator xml:lang="en">Ramdhani, Heparia Arum Ndaru</dc:creator>
	<dc:creator xml:lang="en">Perwira, Rifki Indra</dc:creator>
	<dc:creator xml:lang="en">Nuraini, Dian Meididewi</dc:creator>
	<dc:subject xml:lang="en">Convolutional Neural Network</dc:subject>
	<dc:subject xml:lang="en">EfficientNetV2-B0</dc:subject>
	<dc:subject xml:lang="en">Herbal Leaf Classification</dc:subject>
	<dc:subject xml:lang="en">Learning Media</dc:subject>
	<dc:subject xml:lang="en">Medicinal Plants</dc:subject>
	<dc:description xml:lang="en">The growing use of herbal plants highlights the need for effective learning media that help vocational students identify medicinal plant species more easily. This study aimed to develop an AI-supported, web-based learning tool for medicinal plant leaf identification, evaluate the performance of lightweight Convolutional Neural Network (CNN) models, and assess the tool's feasibility and usability. We used a Research and Development (R&amp;amp;D) approach based on the ADDIE model. We prepared a dataset of 3,100 images across 20 classes to compare four pre-trained lightweight CNN architectures: MobileNetV2, MobileNet, EfficientNet-B0, and EfficientNetV2-B0. The selected model was integrated into the web-based learning tool via a client-server architecture using an Application Programming Interface (API). Model performance was evaluated using accuracy, loss, precision, recall, and F1-score metrics, while feasibility was assessed by media and subject matter experts. 29 vocational students evaluated usability using the System Usability Scale (SUS). EfficientNetV2-B0 achieved the highest average accuracy (98.67%) across three random seeds, and the final model attained 97.25% accuracy on the test set. The tool received feasibility scores of 83.41% from media experts and 96.19% from subject matter experts. The average SUS score was 79.14, indicating good usability from the users' perspective. These findings demonstrate that integrating lightweight CNN models into web-based learning tools can provide a practical resource for learning to identify medicinal plant leaves. The developed tool can support students in both independent and classroom learning, although its broader implementation remains constrained by dataset diversity, device compatibility, internet dependency, and the limited scope of user evaluation.</dc:description>
	<dc:publisher xml:lang="en">LPPM Universitas Jambi</dc:publisher>
	<dc:date>2026-10-31</dc:date>
	<dc:type>info:eu-repo/semantics/article</dc:type>
	<dc:type>info:eu-repo/semantics/publishedVersion</dc:type>
	<dc:format>application/pdf</dc:format>
	<dc:identifier>https://online-journal.unja.ac.id/JIITUJ/article/view/58527</dc:identifier>
	<dc:identifier>10.22437/jiituj.v10i5.58527</dc:identifier>
	<dc:source xml:lang="en">Jurnal Ilmiah Ilmu Terapan Universitas Jambi; Vol. 10 No. 5 (2026): Volume 10, Nomor 5, October 2026; 2524-2535</dc:source>
	<dc:source>2580-2259</dc:source>
	<dc:source>2580-2240</dc:source>
	<dc:source>10.22437/jiituj.v10i5</dc:source>
	<dc:language>en</dc:language>
	<dc:relation>https://online-journal.unja.ac.id/JIITUJ/article/view/58527/25979</dc:relation>
	<dc:rights xml:lang="en">Copyright (c) 2026 Febri Liantoni, Heparia Arum Ndaru Ramdhani, Rifki Indra Perwira, Dian Meididewi Nuraini</dc:rights>
	<dc:rights xml:lang="en">https://creativecommons.org/licenses/by/4.0</dc:rights>
</oai_dc:dc>