Comparative Analysis of Supervised Learning Algorithms on Image Classification of Saouropus Androgynus Domestic Food Processing

Michelle, A. Gregory Qo'nitah (2023) Comparative Analysis of Supervised Learning Algorithms on Image Classification of Saouropus Androgynus Domestic Food Processing. Tugas Akhir (S1) - thesis, Universitas Bakrie.

[thumbnail of Cover]
Preview
Text (Cover)
00. Cover.pdf - Submitted Version

Download (599kB) | Preview
[thumbnail of BAB I-III] Text (BAB I-III)
01. BAB I-III.pdf - Submitted Version
Restricted to Registered users only

Download (9MB)
[thumbnail of BAB IV] Text (BAB IV)
02. BAB IV.pdf - Submitted Version
Restricted to Registered users only

Download (5MB)
[thumbnail of BAB V] Text (BAB V)
03. BAB V.pdf - Submitted Version
Restricted to Registered users only

Download (42kB)
[thumbnail of Daftar Pustaka]
Preview
Text (Daftar Pustaka)
04. BAB Daftar Pustaka.pdf - Submitted Version

Download (88kB) | Preview

Abstract

This research compared supervised machine learning algorithms that classify images. The focus is an image of food processing on katuk leaves in domestic food. Food processing is carried out food processing by boiling and steaming. This research began by collecting a dataset of images captured using a Fujifilm camera and an iPhone camera. After that, the image will be grouping based on the two categories of food processing. Each category will be divided into dataset modification. Before going to the classification stage, the author will preprocess the images that have been collected. In the classification process, the system will classify using supervised machine learning algorithms, namely CNN algorithms. This research applied CNN with few scenario that compared the number epochs, number of layer, and dataset modification. The result of comparing the scenario are the models with the number of epoch 100 have better accuracy than model with the number of epoch 50, the models with a higher number of layers tend to perform better than the ones with fewer layers and The models that use 90% and 80% of the data for training tend to perform slightly better than the ones that use 70%. The final outcome of the alternate scenario is that datasets subjected to cropping processing achieve better accuracy values compared to those that do not undergo the cropping process.

Item Type: Thesis (Tugas Akhir (S1) - )
Uncontrolled Keywords: Food Procesing, Image Classification, Image processing, CNN Algorithm , Confusion Matrix.
Subjects: Computer Science > Image Processing
Thesis > Thesis (S1)
Divisions: Fakultas Teknik dan Ilmu Komputer > Program Studi Informatika
Depositing User: A Gregory Qonitah Michelle
Date Deposited: 25 Aug 2023 09:24
Last Modified: 25 Aug 2023 09:24
URI: https://repository.bakrie.ac.id/id/eprint/8144

Actions (login required)

View Item View Item