Article Per Year (5 Year)
p-Index From 2020 - 2025
0.23
P-Index
This Author published in this journals
All Journal Journal of Embedded Systems, Security and Intelligent Systems
Azir Zuldani Pratama
Unknown Affiliation
Computer Science & IT

Published : 1 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 1 Documents
Search

 1 
Transfer Learning-Based CNN for Guava Fruit Disease Detection and Classification Azir Zuldani Pratama; Mustari Lamada; Surianto, Dewi Fatmarani
Journal of Embedded Systems, Security and Intelligent Systems Vol 6, No 3 (2025): September 2025
Publisher : Program Studi Teknik Komputer

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.59562/jessi.v6i3.10153

Abstract

Guava (Psidium guajava L.) is a tropical plant from the Myrtaceae family and the Psidium genus that is susceptible to diseases such as anthracnose and scab, especially in humid environmental conditions. To accurately detect and classify these diseases, digital image-based technology is needed. However, previous studies still have limitations in dataset size, method variation, and model optimization. Therefore, a study was conducted with the title Guava Fruit Disease Detection and Classification System Using a Convolutional Neural Network (CNN) Based Transfer Learning Model. This study tested four Transfer Learning models, namely MobileNetV2, DenseNet169, VGG16, and EfficientNetV2B5. Based on the test results, the MobileNetV2 model with a combination of activation functions and optimizers (Swish, Swish, Adam) showed the best performance, having the fastest computation time, namely 10 minutes 17 seconds. This proves that the model built is not only superior in accuracy, but also efficient in execution time and can be applied to guava fruit disease detection and classification systems. These findings provide valuable insights into the MobileNetV2 method, combined with Swish, Swish, and Adam, as the best choice for classifying or detecting guava fruit disease levels compared to other methods. This approach can also lead to the development of a widely applicable web-based system for plant disease identification. This offers several benefits for farmers, including faster and more accurate disease detection, efficiency, and cost savings.
Co-Authors Dewi Fatmarani Surianto Mustari Lamada