diplomsko delo
Domen Pogačnik (Avtor), Franc Solina (Mentor)

Povzetek

Ocenjevanje estetike fotografij s pomočjo strojnega učenja

Ključne besede

fotografija;ocenjevanje estetike fotografij;avtomatično ocenjevanje;strojno učenje;računalniški vid;računalništvo;računalništvo in informatika;univerzitetni študij;diplomske naloge;

Podatki

Jezik: Slovenski jezik
Leto izida:
Tipologija: 2.11 - Diplomsko delo
Organizacija: UL FRI - Fakulteta za računalništvo in informatiko
Založnik: [D. Pogačnik]
UDK: 004.85:77(043.2)
COBISS: 9347412 Povezava se bo odprla v novem oknu
Št. ogledov: 49
Št. prenosov: 1
Ocena: 0 (0 glasov)
Metapodatki: JSON JSON-RDF JSON-LD TURTLE N-TRIPLES XML RDFA MICRODATA DC-XML DC-RDF RDF

Ostali podatki

Sekundarni jezik: Angleški jezik
Sekundarni naslov: Evaluating photo aesthetics using machine learning
Sekundarni povzetek: The purpose of this thesis is to identify the characteristics that influence the aesthetic appeal of photographs and to convert these characteristics into calculable features. Primarily, we examined the relation between the foreground object - subject and the photograph as a whole. The subject was manually identified with the assistance of experienced photographers. During our research, we implemented 31 features to analyze various aspects of a photograph, such as the color scheme, composition and proportions. A simple web application for labeling and the aesthetic assessment of photographs, which was developed as a part of our thesis, was used by experienced photographers in order to provide learning samples. On the basis of calculated features and gathered learning data, we used machine learning algorithms to create a model which is able to distinguish high quality/professional from low quality/snapshot photographs. We achieved 93 percent classification accuracy using SVM classifier. The features used by machine learning algorithm were analyzed with reliefF metric and the nomogram data provided by the Naive Bayes classifier. In the final part, we presented and discussed the influence of calculated features and suggested some guidelines for further research on the subject.
Sekundarne ključne besede: photo aesthetics;evaluating photo aesthetics;machine learning
Vrsta datoteke: application/pdf
Vrsta dela (COBISS): Diplomsko delo/naloga
Komentar na gradivo: Univ. v Ljubljani, Fak. za računalništvo in informatiko
Strani: 65 str.
ID: 24142517