Febby Surya Ayu Nengsih, Febby Surya Ayu Nengsih and Untung Suwardoyo, Untung Suwardoyo and A. Irmayani P, A. Irmayani P and Masnur, Masnur (2025) RANCANG BANGUN APLIKASI PENDETEKSI TINGKAT KEKERINGAN PADA BIJI KAKAO BERBASIS ANDROID. Jurnal Manajemen dan Teknologi Informasi (JMTI), 15 (2). pp. 39-49. ISSN 2721-7604
RANCANG BANGUN APLIKASI PENDETEKSI TINGKAT.pdf - Published Version
Download (921kB) | Preview
Abstract
The cocoa industry faces challenges in ensuring the quality of cocoa beans, especially in detecting the level of dryness that affects the taste and quality of the final product. This study aims to design and develop an Android-based application that can detect the level of dryness of cocoa beans using machine learning technology, specifically Convolutional Neural Network (CNN) implemented with TensorFlow Lite. This application aims to provide a practical and efficient solution for farmers, especially in remote areas, using easily accessible Android devices. The CNN model was trained using a dataset of cocoa bean images with three categories of dryness: Wet, Half Dry, and Dry. The application was tested on 30 respondents consisting of farmers and cocoa bean processors to assess detection accuracy and ease of use. The test results showed that the application was able to detect the level of dryness of cocoa beans with adequate accuracy and appropriate detection speed on Android devices with low specifications. Although this application shows significant potential, several challenges remain, especially related to optimizing detection in poor lighting conditions. This research contributes to the use of mobile technology and machine learning to improve the quality of agricultural products, as well as opening up opportunities for further development in digital-based agricultural systems.
| Item Type: | Article |
|---|---|
| Subjects: | Sains dan Teknologi |
| Depositing User: | Irmayani |
| Date Deposited: | 07 Oct 2026 00:25 |
| Last Modified: | 07 Oct 2026 00:25 |
| URI: | https://repository.umpar.ac.id/id/eprint/213 |
