Sistem Pengusir Hama Burung Pada Tanaman Padi Berbasis Kamera dan Algoritma YOLO

Yogi, Muhamad (2026) Sistem Pengusir Hama Burung Pada Tanaman Padi Berbasis Kamera dan Algoritma YOLO. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5022221039_Undergraduate-Thesis.pdf Restricted to Repository staff only Download (7MB) | Request a copy

Abstract

Burung pipit (Lonchura punctulata) merupakan organisme pengganggu tanaman utama pada fase generatif padi yang berpotensi memicu kerugian panen secara signifikan. Metode pengusiran konvensional rentan terhadap fenomena habituasi, sehingga efektivitas pencegahannya cenderung berumur pendek. Penelitian ini bertujuan merancang perangkat sistem pengusir hama cerdas berbasis visi komputer dan aktuator audio adaptif menggunakan perangkat komputasi edge Raspberry Pi 4 Model B. Sistem deteksi mengimplementasikan algoritma YOLOv8 untuk pengenalan target secara real-time, yang dipadukan dengan model Pinhole Camera untuk mengekstraksi jarak dan sudut posisi objek burung pipit. Intensitas output suara dikendalikan secara dinamis menggunakan potensiometer digital X9C103 dan diperkuat oleh penguat daya kelas D TPA3116D2 untuk menghasilkan efek sebaran suara spasial pada rentang sensitivitas pendengaran hama, yakni 2-8 kHz. Hasil evaluasi menunjukkan bahwa model deteksi mampu mencapai nilai Precision sebesar 86% dan Recall sebesar 83% tanpa adanya error klasifikasi silang. Sistem penglihatan komputer juga mencatatkan penyimpangan deviasi sudut yang rendah, yakni antara 4,0° hingga 6,84° pada rentang jarak operasional 2-7 meter. Pada kendali akustik, rangkaian atenuasi digital beroperasi dengan tingkat error rata-rata 10,75%, sementara modul penguat daya berhasil melipatgandakan amplitudo output dengan rata-rata 17,8 kali penguatan hingga titik maksimal 12,8 Vpp. Secara keseluruhan, sistem mampu memfokuskan tembakan gelombang suara dari sudut -40° hingga +40° secara dinamis mengikuti koordinat target. Kendati pengujian pada area persawahan menunjukkan adanya batasan deteksi visual akibat halangan fisik tanaman padi, uji habituasi komparatif di lapangan membuktikan bahwa stimulus suara burung elang merupakan output yang paling efektif dan tangguh terhadap adaptasi hama dibandingkan suara lain. Secara keseluruhan, keandalan instrumen aktuator akustik dan keakuratan lokalisasi target mengonfirmasi bahwa arsitektur sistem ini secara fungsional siap dikembangkan sebagai solusi mitigasi hama pertanian yang adaptif.
=====================================================================================================================================
The sparrow (Lonchura punctulata) is a major agricultural pest during the generative phase of rice crops, potentially causing significant yield losses. Conventional repelling methods are susceptible to habituation, making their preventive effectiveness tend to be short-lived. This research aims to design a smart pest repellent system based on computer vision and adaptive audio actuators utilizing a Raspberry Pi 4 Model B edge computing device. The detection system implements the YOLOv8 algorithm for real-time target recognition, integrated with a Pinhole Camera model to extract the distance and angular position of the sparrow. The sound output intensity is dynamically controlled using an X9C103 digital potentiometer and amplified by a TPA3116D2 Class-D power amplifier to produce a spatial sound panning effect within the pest's hearing sensitivity range of 2-8 kHz. Evaluation results indicate that the detection model achieved a Precision of 86% and a Recall of 83% without any cross-classification errors. The computer vision system also recorded a low angular deviation, ranging from 4.0° to 6.84° within an operational distance of 2-7 meters. In the acoustic control subsystem, the digital attenuation circuit operated with an average error rate of 10.75%, while the power amplifier module successfully boosted the output amplitude with an average gain of 17.8 times, reaching a maximum of 12.8 Vpp. Overall, the system is capable of dynamically directing focused sound waves from an angle of -40° to +40° following the target's coordinates. Although field testing in paddy fields revealed visual detection limitations due to physical obstructions by rice plants, the comparative field habituation test proved that the eagle sound stimulu is the most effective and robust output against pest adaptation compared to other sounds. Overall, the reliability of the acoustic actuator instruments and the accuracy of target localization confirm that this system architecture is functionally viable for development as an adaptive agricultural pest mitigation solution.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Burung Pipit, Pinhole Camera, Raspberry Pi, YOLOv8, Audio Adaptif. Sparrow, Pinhole Camera, Raspberry Pi, YOLOv8, Adaptive Audio.
Subjects:	Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines. Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science) S Agriculture > S Agriculture (General)
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Electrical Engineering > 20201-(S1) Undergraduate Thesis
Depositing User:	Muhamad Yogi
Date Deposited:	22 Jul 2026 03:11
Last Modified:	22 Jul 2026 03:11
URI:	http://repository.its.ac.id/id/eprint/136143

Actions (login required)

View Item