LarvaDetect.ai: Pengembangan Aplikasi Mobile Berbasis Android untuk Identifikasi Spesies Larva Nyamuk dengan Implementasi Deep Learning

Rafliansyah, Sulthan Akmal (2025) LarvaDetect.ai: Pengembangan Aplikasi Mobile Berbasis Android untuk Identifikasi Spesies Larva Nyamuk dengan Implementasi Deep Learning. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5027211039-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (9MB) | Request a copy

Abstract

Nyamuk merupakan vektor penyebar berbagai penyakit menular seperti demam berdarah, malaria, dan filariasis. Identifikasi larva nyamuk merupakan langkah penting dalam pengendalian populasi nyamuk, namun proses manual identifikasi spesies larva membutuhkan waktu dan keahlian khusus. Penelitian ini bertujuan untuk mengembangkan sistem klasifikasi larva nyamuk berbasis deep learning dengan pendekatan transfer learning yang dapat diimplementasikan dalam aplikasi mobile Android. Lima arsitektur Convolutional Neural Network (CNN) yaitu ResNet50, VGG16, MobileNetV2, EfficientNetB0, dan InceptionV3 digunakan untuk melatih model berdasarkan empat bagian tubuh larva: kepala, abdomen ke-8, siphon, dan seluruh tubuh. Dataset berupa citra mikroskopis larva diperoleh dari Laboratorium Entomologi Universitas Airlangga dan dilakukan preprocessing serta augmentasi. Hasil evaluasi menunjukkan bahwa model EfficientNetB0 memberikan hasil terbaik pada bagian tubuh dengan struktur besar seperti abdomen dan siphon, sementara InceptionV3 unggul dalam mengenali bagian kecil seperti kepala dan full body. Akurasi evaluasi model pada sebagian besar kombinasi model dan bagian tubuh mencapai 97,5% hingga 100%, meskipun terdapat indikasi overfitting pada grafik pelatihan. Model terbaik kemudian dikonversi ke format TensorFlow Lite dan diintegrasikan ke dalam aplikasi LarvaDetect.ai. Aplikasi diuji menggunakan 135 gambar larva dalam tiga variasi pencahayaan (gelap, normal, terang), dan menghasilkan akurasi klasifikasi keseluruhan sebesar 94,07%, dengan performa terbaik pada pencahayaan normal (95,56%). Sistem ini terbukti mampu melakukan klasifikasi larva nyamuk secara cepat dan akurat langsung dari perangkat seluler, sehingga berpotensi besar untuk membantu identifikasi larva baik di laboratorium maupun lapangan. =============================================================================================================================================
Mosquitoes are vectors of various infectious diseases such as dengue fever, malaria, and filariasis. Identifying mosquito larvae is a crucial step in controlling their population; however, manual species identification requires considerable time and specific expertise. This study aims to develop a mosquito larvae classification system based on deep learning using a transfer learning approach, which is implemented into an Android mobile application. Five Convolutional Neural Network (CNN) architectures—ResNet50, VGG16, MobileNetV2, EfficientNetB0, and InceptionV3—were used to train models using four larval body parts: the head, 8th abdominal segment, siphon, and full body. The dataset comprised microscopic images of larvae collected from the Entomology Laboratory of Universitas Airlangga and was subjected to preprocessing and augmentation. Evaluation results show that EfficientNetB0 performed best on larger structures such as the abdomen and siphon, while InceptionV3 excelled in recognizing smaller structures like the head and full body. Most model evaluations achieved accuracy between 97.5% and 100%, although signs of overfitting were observed during training visualization. The best-performing model was converted into TensorFlow Lite format and integrated into the LarvaDetect.ai application. The system was tested using 135 larval images with three brightness variations (dark, normal, bright), resulting in an overall classification accuracy of 94.07%, with the highest performance under normal lighting conditions (95.56%). The system demonstrates the ability to accurately and efficiently classify mosquito larvae directly from mobile devices, offering strong potential for use in both laboratory and field environments.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Identifikasi Larva Nyamuk, Klasifikasi Larva Nyamuk, Deep Learning, Convolutional Neural Network, Mosquito Larvae Identification, Mosquito Larvae Classification, Deep Learning, Convolutional Neural Network.
Subjects:	T Technology > T Technology (General)
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Information Technology > 59201-(S1) Undergraduate Thesis
Depositing User:	Sulthan Akmal Rafliansyah
Date Deposited:	14 Jul 2025 01:58
Last Modified:	14 Jul 2025 02:01
URI:	http://repository.its.ac.id/id/eprint/119556

Actions (login required)

View Item