APLIKASI MONITORING PANEN DAN IDENTIFIKASI KEMATANGAN BUAH KELAPA SAWIT BERBASIS ANDROID

Abstract

Abstract: The development of information technology has contributed to increasing the efficiency of plantation sector management, particularly in the process of harvest monitoring and identification of oil palm fruit ripeness. Determining the level of fruit ripeness is still done manually, potentially causing inconsistencies due to differences in perception among field officers. This study aims to design and implement an Android-based application for Harvest Monitoring and Identification of Oil Palm Fruit Maturity that can assist the process of recording harvest results, monitoring harvest activities, and digitally identifying fruit ripeness levels. The system was developed through the stages of data collection, image pre-processing, ripeness identification using machine learning or deep learning methods, data storage, and information presentation through an Android application. The main features of the application include ripeness identification, detection history, location maps, harvest schedules, harvest monitoring, and harvest reports. The evaluation system was carried out using Black Box Testing to test the application's functionality, while model performance was evaluated based on training accuracy and testing accuracy values. The results showed that the model training process ran well without any indication of significant overfitting. Furthermore, the test results yielded accuracy values ​​ranging from 94.8% to 97.1%, with an average above 95%, indicating that the model has good generalization capabilities in identifying oil palm fruit ripeness. The developed application can support more effective harvest monitoring, improve the accuracy of fruit ripeness identification, and facilitate integrated harvest data management via Android devices.

Keywords: harvest monitoring, ripeness identification, oil palm, Android, machine learning, deep learning.

Abstrak: Perkembangan teknologi informasi telah memberikan kontribusi dalam meningkatkan efisiensi pengelolaan sektor perkebunan, khususnya pada proses monitoring panen dan identifikasi kematangan buah kelapa sawit.. Penelitian ini bertujuan untuk merancang dan mengimplementasikan aplikasi Monitoring Panen dan Identifikasi Kematangan Buah Kelapa Sawit Berbasis Android yang mampu membantu proses pencatatan hasil panen, monitoring aktivitas panen, serta identifikasi tingkat kematangan buah secara digital. Sistem dikembangkan melalui tahapan pengumpulan data, pra-pemrosesan citra, identifikasi kematangan menggunakan metode pembelajaran mesin (machine learning) atau deep learning, penyimpanan data, dan penyajian informasi melalui aplikasi Android. Fitur utama aplikasi meliputi identifikasi kematangan, riwayat deteksi, peta lokasi, jadwal panen, monitoring panen, dan laporan hasil panen. performa model dievaluasi berdasarkan nilai training accuracy dan testing accuracy. Hasil penelitian menunjukkan bahwa proses pelatihan model berjalan dengan baik tanpa indikasi overfitting yang signifikan. Selain itu, hasil pengujian memperoleh nilai testing accuracy pada rentang 94,8% hingga 97,1% dengan rata-rata di atas 95%, yang menunjukkan bahwa model memiliki kemampuan generalisasi yang baik dalam mengidentifikasi tingkat kematangan buah kelapa sawit. Aplikasi yang dikembangkan mampu mendukung proses monitoring panen secara lebih efektif, meningkatkan akurasi identifikasi kematangan buah, serta mempermudah pengelolaan data panen secara terintegrasi melalui perangkat Android.

Kata kunci: monitoring panen, identifikasi kematangan, kelapa sawit, Android, machine learning, deep learning.