Classification Fashion Image Using Local Binary Pattern And Artificial Neural Network Multi Layer Perceptron

Abstract

Abstrak Busana merupakan kebutuhan pokok yang selalu manusia pakai setiap hari. Pakaian digunakan manusia sebagai pelindung dan sebagai penutup tubuh. Namun seiring perkembangan kehidupan, pakain dijadikansimbolkedudukan,status,jabatandansebagaipakainuntukberaktivitas. Karenanyajenisbusana bermacam-macamdimulaidaripakaianuntuktidursampaipakainuntukbekerja. Sistemklasiï¬kasidibutuhkan dalam untuk bisa mengkategorikan busana terutama bagi e-commerce dan lain-lain. Pada sistem yang kami rancang, Terdiri dari tiga tahapan antara lain: Preprocessing, Fitur Ekstraksi dan Klasiï¬kasi. Pada tahap Preprocessing data akan di resize kemudian dikonversi menjadi grayscale. Kemudian pada tahapFiturekstraksigambarakandidiprosesdenganLBPatauLocalBinaryPatternsetelahitucitraakan di menjadi 2 yaitu citra training dan citra untuk testing. Citra untuk Training akan dimasukan terhadap modelJST,modelJSTakanbelajardaridatatraininguntukmenemukanpolapadadatatersebut. Setelah tahapan training selesai citra akan di tes untuk mengetahui akurasi terhadap data yang belum pernah di dipelajari dengan menggunakandata test sebagai pengujiancitra yang akan diidentiï¬kasi. Mencapai nilai akurasi96.38%
Kata kunci :Image Processing, Jaringan Saraf Tiruan, Classiï¬cation, Fashion Classiï¬cation, Local Binary Pattern
Abstract Clothing Fashion is a basic need that humans always use every day. Clothing is used by humans as a protector and as a body covering. But along with the development of life, Fashion are used as symbols of position, status, position and as clothes for activities. Therefore various types of clothing ranging from clothing for sleeping to clothes for work. A classiï¬cation system is needed in order to be able to categorize clothing especially for e-commerce and others. In the system that we designed, it consists of three stages including: Preprocessing,FeatureExtractionandClassiï¬cation. AtthePreprocessingstagethedatawillbe resizedandthenconvertedtograyscale. ThenintheimageextractionfeaturewillbeprocessedwithLBPor Local Binary Pattern after that the image will be in 2, namely training images and images for testing. The image for the training will be added to the ANN model, the ANN model will learn from the training data to ï¬nd patterns in the data. After the training stage is completed the image will be tested to determine the accuracy of the data that has never been studied by using test data as an image test that will be identiï¬ed. Reach96.38%accuracy.
Keywords: Image Processing, Artiï¬cial Neural Network, Classiï¬cation, Clothing Classiï¬cation, Local BinaryPattern