Penilaian Estetika Foto Makanan Berdasarkan Komposisi Visual

Dianto, Riki Wahyu Nur (2023) Penilaian Estetika Foto Makanan Berdasarkan Komposisi Visual. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 05111940000188-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2025. Download (5MB) | Request a copy

Abstract

Makanan merupakan salah satu aspek terpenting dalam kehidupan manusia. Foto makanan dapat memengaruhi persepsi dan keinginan manusia terhadap makanan. Saat ini di media sosial terdapat lebih dari enam ratus juta pengguna aktif yang mengunggah foto mereka termasuk foto makanan di dalamnya. Adanya platform yang dapat menilai estetika dari foto tersebut tentu akan sangat membantu khususnya pada bidang yang mementingkan kualitas estetika visual dari suatu foto, contohnya seperti memudahkan fotografer maupun pemilik bisnis di industri makanan mendapat foto terbaik untuk kebutuhan mereka. Untuk menilai estetika sebuah foto, komposisi visual (seperti posisi dan ukuran elemen visual) merupakan merupakan salah satu aspek penting yang harus diperhatikan. Pola komposisi yang berbeda menawarkan perspektif yang berbeda dalam evaluasi kualitas komposisi. Selain itu, salah satu elemen komposisi yang paling penting adalah warna. Warna dapat digunakan untuk mengarahkan perhatian pada elemen-elemen tertentu dalam foto dan memengaruhi suasana hati para pelihatnya. Dimotivasi oleh hal tersebut, Tugas Akhir ini mengeksplorasi komposisi visual foto makanan untuk mengevaluasi nilai estetikanya. Komposisi visual foto diekstraksi dan digunakan sebagai fitur input untuk proses pelatihan model machine learning. Model ini selanjutnya digunakan dalam memprediksi estetika foto makanan. Eksperimen dilakukan dengan memanfaatkan sebuah benchmark dataset Gourmet Photography Dataset. Performa model akan dievaluasi berdasarkan nilai akurasi, presisi, recall, dan f1-score. Dalam Tugas Akhir ini menggunakan model Support Vector Machine (SVM), Decision Tree, Random Forest, Multilayer Perceptron (MLP), Hybrid Inception V3 + MLP, dan Hybrid VGG-16 + MLP. Dari keenam model tersebut, model Hybrid VGG-16 + MLP memiliki performa terbaik. Model Hybrid VGG-16 + MLP memiliki akurasi 0,912, presisi 0,910, recall 0,932, dan f1-score 0,921 membuat memiliki performa terbaik jika dibandingkan dengan model lain yang digunakan dalam Tugas Akhir ini.
=================================================================================================================================
Food is one of the most important aspects of human life. Food photos can influence people's perception and desire for food. Currently on social media, there are more than six hundred million active users who upload their photos including food photos. The existence of a platform that can assess the aesthetics of these photos will certainly be very helpful, especially in fields that are concerned with the visual aesthetic quality of a photo, such as making it easier for photographers and business owners in the food industry to get the best photos for their needs. To assess the aesthetics of a photograph, visual composition (such as the position and size of visual elements) is one of the important aspects to be considered. Different composition patterns offer different perspectives in the evaluation of composition quality. Also, one of the most important elements of composition is color. Color can be used to direct attention to certain elements in a photo and influence the mood of the viewer. Motivated by that, this Final Project explores the visual composition of food photos to evaluate their aesthetic value. The visual composition of the photos is extracted and used as input features for the machine learning model training process. This model is then used in predicting the aesthetics of food photos. Experiments are conducted by utilizing a benchmark dataset Gourmet Photography Dataset. The performance of the model will be evaluated based on accuracy, precision, recall, and f1-score values. In this Final Project, Support Vector Machine (SVM), Decision Tree, Random Forest, Multilayer Perceptron (MLP), Hybrid Inception V3 + MLP, and Hybrid VGG-16 + MLP models are used. Of the six models, the Hybrid VGG-16 + MLP model has the best performance. The Hybrid VGG-16 + MLP model has an accuracy of 0.912, precision of 0.910, recall of 0.932, and f1-score of 0.921 making it the best performance when compared to other models used in this Final Project.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Estetika, Foto Makanan, Komposisi Visual, Machine Learning, Aesthetic, Food Photos, Gourmet Photography Dataset, Visual Composition.
Subjects:	Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines. T Technology > T Technology (General) > T58.6 Management information systems T Technology > TR Photography
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Informatics Engineering > 55201-(S1) Undergraduate Thesis
Depositing User:	Riki Wahyu Nur Dianto
Date Deposited:	13 Nov 2023 04:19
Last Modified:	13 Nov 2023 04:19
URI:	http://repository.its.ac.id/id/eprint/102596

Actions (login required)

View Item