Deteksi Suara Burung Pipit Menggunakan Convolutional Neural Network

Acalapati, Moch. Ardaffa Tsaqif (2024) Deteksi Suara Burung Pipit Menggunakan Convolutional Neural Network. Other thesis, Institut Teknologi Sepuluh Nopember.

[thumbnail of 02311940000149_Undergraduate_Thesis.pdf] Text
02311940000149_Undergraduate_Thesis.pdf - Accepted Version
Restricted to Repository staff only until 1 April 2026.

Download (3MB) | Request a copy

Abstract

Pemantauan hewan penting untuk pelestarian dan konservasi lingkungan. Hal tersebut memungkinkan peneliti dan aktivis lingkungan untuk menilai dampak aktivitas di bumi. Di antara semua hewan, Burung adalah salah satu hewan yang paling sering dipantau karena perubahan populasi burung dapat mengindikasikan adanya perubahan pada ekosistem di wilayah tersebut, khususnya burung pipit. Burung pipit memainkan peran penting dalam ekosistem dengan mengontrol hama dan serangga serta berfungsi sebagai sumber makanan bagi hewan lain. Namun, Keberadaan burung pipit seringkali menjadi ancaman besar bagi petani padi sawah. Petani sering menggunakan cara-cara seperti membuat orang-orangan sawah atau menggantung tali dengan kaleng bekas untuk mengusir hama burung pada tanaman padi. Namun, jika tidak berhasil, mereka harus turun langsung ke sawah untuk mengusir burung tersebut. Beberapa petani mempekerjakan orang untuk menjaga sawah, meskipun ini tidak efektif dan efisien karena biaya upahnya. Oleh karena itu, diperlukan pengendalian hama burung yang lebih efektif dan efisien. Pemantauan burung yang tergolong sulit karena penyebaran burung yang luas dan keadaan lingkungan dengan rendah pencahayaan mendorong pemantauan dengan suara daripada visual.Pada penelitian ini digunakan model convolutional neural network (CNN) sederhana untuk mendeteksi suara burung dengan menggunakan input suara yang sudah diubah menjadi gambar spektogram. Dataset yang digunakan terbagi menjadi dua yaitu rekaman langsung dan dataset online xeno-canto. Hasil akurasi yang didapatkan dari proses training dan validasi menggunakan CNN sederhana didapatkan sebesar 98 %.
=================================================================================================================================
Animal monitoring is important for environmental preservation and conservation. This allows researchers and environmental activists to assess the impact of activities on earth. Among all animals, birds are one of the most frequently monitored animals because changes in bird populations can indicate changes in the ecosystem of the region, particularly sparrows. Sparrows play an important role in the ecosystem by controlling pests and insects and serving as a food source for other animals. However, the existence of sparrows is often a big threat to rice farmers. Farmers often use methods such as making scarecrows or hanging ropes with old cans to repel bird pests on rice crops. However, if it doesn't work, they have to go down directly to the rice fields to drive the bird away. Some farmers hire people to guard the fields, although this is not effective and efficient due to the cost of wages. Therefore, more effective and efficient bird pest control is needed. Bird monitoring is difficult because the wide spread of birds and low-lighting environments encourage sound rather than visual monitoring. In this study, a simple convolutional neural network (CNN) model was used to detect bird sounds using sound inputs that have been converted into spectrogram images. The dataset used is divided into two, namely live recording and xeno-canto online dataset. The accuracy results obtained from the training and validation process using a simple CNN were obtained at 98%.

Item Type: Thesis (Other)
Uncontrolled Keywords: Burung Pipit, Spektogram, Sparrow, Convolutional Neural Network, Spectrogram
Subjects: T Technology > T Technology (General) > T57.5 Data Processing
T Technology > T Technology (General) > T58.5 Information technology. IT--Auditing
T Technology > T Technology (General) > T59.7 Human-machine systems.
Divisions: Faculty of Industrial Technology and Systems Engineering (INDSYS) > Physics Engineering > 30201-(S1) Undergraduate Thesis
Depositing User: Moch. Ardaffa Tsaqif Acalapati
Date Deposited: 15 Mar 2024 01:52
Last Modified: 15 Mar 2024 01:52
URI: http://repository.its.ac.id/id/eprint/107809

Actions (login required)

View Item View Item