Estimasi Bobot Sapi Berdasarkan Citra Digital Dengan Metode Fraktal Dan Klasifikasi Decision Tree

Fajar Kurniawan Alhamal, Jangkung Raharjo, Syamsul Rizal

Abstract

Abstrak
Potensi sapi di Indonesia mempunyai peluang yang sangat besar, karena jenis sapi
endemik Indonesia masuk jajaran sapi paling berkualitas di dunia. Dalam menentukan
kualitas sapi, bobot merupakan salah satu indikator penting. Dengan bobot, peternak dapat
menentukan hasil produksi dan produktivitas sapi. Menentukan bobot sapi yang paling
umum adalah menggunakan timbangan. Namun, mahalnya harga timbangan menjadi salah
satu faktor penghambat dalam merintis usaha peternakan sapi.
Sistem yang dirancang menggunakan masukan citra sapi dari sisi samping dan
keluaran berupa estimasi bobot sapi. Tujuan dari penelitian ini adalah mempermudah calon
peternak dalam menentukan bobot sapi tanpa menggunakan timbangan yang harganya
relatif mahal.
Sistem yang telah dirancang dalam program aplikasi estimasi bobot sapi memerlukan
input berupa citra atau gambar sapi dan menghasilkan output berupa bobot beserta
klasifikasi sapi berdasarkan bobot sapi yang diperoleh. Program aplikasi yang
diimplementasikan untuk mengestimasi bobot sapi, dirancang dalam software MATLAB
2018a menggunakan metode fraktal dan klasifikasi Decision Tree. Pada tugas akhir ini
mendapatkan tingkat akurasi estimasi sistem sebesar 81% dengan nilai root mean squared
error pada perhitungan rumus schoorl mendapatkan hasil 72,56277, winter 75,00148, dan
denmark 69,11267. Waktu komputasi rata-rata 0,3329 detik. Akurasi dan waktu komputasi
didapatkan dengan jumlah data latih sebanyak 47 citra dan jumlah data uji sebanyak 21 citra.
Kata kunci: Bobot Sapi, Decision Tree, Fraktal, Pengolahan Citra Digital.
Abstract
Potential cattle in Indonesia has a very big opportunity, because Indonesian endemic cows
enter ranks of the most qualified cows in the world. In determining the quality of the cow, weight
is an important indicator. With weight, breeders can determine cow production and productivity.
Determines the most common cow weight is to use the scales. However, the high price of the scales
is wrong one inhibiting factor in starting a cattle farm business.
System which is designed using a cow image input from the side and output in the form of
cow weight estimation. The purpose of this research is to make it easier for candidates breeders
in determining the weight of the cow without using scales the price is relatively expensive.
The system that has been designed in the cow weight estimation application program
requires input in the form of an image or image of a cow and produces output in the form weight
and classification of cattle based on the weight of cattle obtained. Program the application
implemented to estimate the weight of a cow is designed in MATLAB 2018a software uses the
fractal method and the Decision classification Tree. In this final project, the system estimation
accuracy rate is 81% with the root mean squared error value in the Schoorl formula calculation,
getting 72.56277 results, winter 75.00148, and Denmark 69.11267. The average computation time
was 0.3329 seconds. Accuracy and computation time obtained by the amount of training data as
much as 47 images and the number of test data as many as 21 images.
Key words : Fractal, Decision Tree, Cow Weight, Digital Image Processing

Full Text:

PDF

Refbacks

  • There are currently no refbacks.
max_upload :0