Klasifikasi Objek Pariwisata Melalui Unggahan Foto di Media Sosial dengan Metode Pre-trained Convolutional Neural Network

Ilmi, Akhmad Miftakhul (2023) Klasifikasi Objek Pariwisata Melalui Unggahan Foto di Media Sosial dengan Metode Pre-trained Convolutional Neural Network. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 06211940000062_Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2025. Download (4MB) | Request a copy

Abstract

Pariwisata merupakan salah satu sektor penting dalam aktivitas sosial dan ekonomi suatu negara. Saat ini sektor pariwisata erat kaitannya dengan media sosial, karena banyak wisatawan yang mengunggah foto berwisata maupun ulasan tempat wisata di akun sosial medianya. Promosi yang tepat sasaran pada media sosial merupakan salah satu strategi untuk mengembangkan sektor pariwisata, oleh karena itu perlu dibangun sistem rekomendasi objek pariwisata. Dalam membangun sistem rekomendasi khususnya yang berbasis Fuzzy Rules, umumnya perlu dilakukan klasifikasi objek pariwisata terlebih dahulu. Maka dari itu, pada penelitian ini akan dilakukan klasifikasi citra foto dari objek pariwisata menggunakan tiga jenis arsitektur pre-trained model Convolutional Neural Network (CNN) yaitu GoogLeNet, AlexNet, dan VGG16. Training pada 6000 data dilakukan pada ketiga jenis arsitektur pre-trained model CNN tersebut. Kemudian, masing-masing pre-trained model dilakukan training sebanyak lima kali, dengan kombinasi hyperparameter berbeda-beda untuk menemukan performa terbaik dari masing-masing model. Setelah dilakukan perbandingan, didapatkan model terbaik yaitu GoogLeNet dengan kombinasi hyperparameter meliputi penggunaan layer inception block 5 dalam kondisi unfreeze, learning rate sebesar 0,0001, batch size 32, dan dropout 0,5. Model dengan kombinasi tersebut mencapai akurasi data training sebesar 0,9871, nilai loss data training sebesar 0,0631, akurasi data validation sebesar 0,9125, dan nilai loss data validation sebesar 0,2533. Selanjutnya model tersebut diaplikasikan pada 600 data testing dan menghasilkan akurasi sebesar 0,9467 dan nilai loss sebesar 0,1158. Model ini memiliki performa yang baik dalam mengklasifikasikan objek pariwisata pada data test, termasuk dalam konteks objek pariwisata di Indonesia.
=================================================================================================================================
Tourism is an important sector in the social and economic activities of a country. Currently, the tourism sector is closely related to social media, as many tourists upload photos of tourist destinations on their social media accounts. Targeted promotion on social media is one of the strategies to develop the tourism sector, therefore it is necessary to build a tourism object recommendation system. In building recommendation systems, especially those based on Fuzzy Rules, it is generally necessary to classify tourism objects first. Therefore, this research will classify image photos of tourism objects using three types of pre-trained Convolutional Neural Network (CNN) models: GoogLeNet, AlexNet, and VGG16. Training will be conducted on 6000 data using these three types of pre-trained CNN models. Then, each pre-trained model will be trained five times with different combinations of hyperparameters to find the best performance for each model. After comparison, the best model is found to be GoogLeNet with the combination of hyperparameters including the use of the inception block 5 layer in an unfreeze condition, a learning rate of 0.0001, a batch size of 32, and a dropout of 0.5. The model with this combination achieves a train accuracy of 0.9871, a train loss of 0.0631, a validation accuracy of 0.9125, and a validation loss of 0.2533. Subsequently, this model is applied to 600 testing data and achieves an accuracy of 0.9467 and loss of 0,1158. This model performs well in classifying tourism objects in the test data, including in the context of tourism objects in Indonesia.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Classification, CNN, Pre-trained model, Tourism, Klasifikasi, Pariwisata
Subjects:	Q Science > Q Science (General) > Q325.5 Machine learning. Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science)
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49201-(S1) Undergraduate Thesis
Depositing User:	Akhmad Miftakhul Ilmi
Date Deposited:	02 Oct 2023 08:01
Last Modified:	02 Oct 2023 08:01
URI:	http://repository.its.ac.id/id/eprint/104546

Actions (login required)

View Item