Analisis Metode Deep Learning dalam Kategorisasi Produk Secara Otomatis

Muhammad David, Hilmawan (2022) Analisis Metode Deep Learning dalam Kategorisasi Produk Secara Otomatis. Undergraduate Thesis thesis, Institut Teknologi Telkom Purwokerto.

	Text Cover.pdf Download (4MB)
	Text Abstract.pdf Download (131kB)
	Text Abstrak.pdf Download (113kB)
	Text BAB I.pdf Download (244kB)
	Text BAB II.pdf Download (494kB)
	Text BAB III.pdf Download (499kB)
	Text BAB IV.pdf Restricted to Registered users only Download (587kB) | Request a copy
	Text BAB V.pdf Download (112kB)
	Text Daftar Pustaka.pdf Download (192kB)
	Text Lampiran.pdf Restricted to Registered users only Download (1MB) | Request a copy

Abstract

At the end of the year 2019, in Wuhan City, Hubei Province, China, was found the world's first Covid-19 cases. This disease was caused by the Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) virus which can spread quickly and easily through the air. this has caused Covid-19 to become a global pandemic that has made most countries carry out lockdown procedures that forced almost all their citizens to stay at home and rely on online orders to buy daily necessities. This has made many physical stores change their sales model to online sales through the marketplace to earn income during the Covid-19 pandemic. Many shop owners use the marketplace for the first time to sell their products. The store owner's lack of familiarity with using marketplace can result in errors such as entering their product to the wrong category. This can be minimized by creating a system that uses deep learning to check the categories of products that the sellers upload automatically. In this research, the model that will be used to classify the category of a product from the images of the product that the seller upload using deep learning method convolutional neural network algorithm with state-of-the-art EfficientNet architecture. this research was made using product images from 42 product categories that was obtained from Shopee Online Shop. The step of preprocessing data for this research starts with data preprocessing, splitting data into training data for training the model and testing data to evaluate the performance of the model using confusion matrix. The model achieved results with precision of 87,41%, recall of 85,10%, accuracy of 85,73%, and f1-score of 86,24%. The implementation results in predicting the Mask R-CNN segmentation results get an average of 20 seconds in the process. Keyword: categorization, automated, marketplace, deep learning, convolutional neural network, efficientnet

Actions (login required)

View Item