APLIKASI MOBILE DETEKSI PENYAKIT DEMODEKOSIS PADA ANJING MENGGUNAKAN METODE CONVOLUTIONAL NEURAL NETWORK

Abstract

Demodex is a skin disease in dogs caused by the protozoan mite Demodex
follicularumvar canis or Demodex canis. These mites are found in the hair roots and sometimes
in the glands, almost the entire life cycle of this type of mite is on the skin, if it lasts for a long
time it can cause death in dogs that are infested by these lice This study aims to detect
demodicosis in dogs with a mobile application using a convolutional neural network method
with a squeezenet architecture model with radam optimization. This application works by taking
pictures directly through the cellphone camera then processing it with a trained model then the
output of this application displays the probability value of the dog being indicated as demodex
or not. Based on the results of testing using the black box testing method, the overall
functionality of the application has been declared successful in accordance with its respective
functions and objectives. From the results of validation data testing using 60 validation data,
based on the accuracy demodecosis detection application is able to recognize validation data
correctly as many as 45 data and 15 data has not been properly recognized, so that the
accuracy value is 75% and the f1-score value is 0,595.