Υπολογιστικό σύστημα ταξινόμησης έργων ζωγραφικής με την χρήση τεχνητών νευρωνικών δικτύων

In this thesis, we studied the problem of automated classification of style in artistic paintings. Our aim was to design a system for the classification of paintings according to artistic style, using machine learning algorithms. From the field of machine learning we focused on artificial neural...

Πλήρης περιγραφή

Αποθηκεύτηκε σε:
Λεπτομέρειες βιβλιογραφικής εγγραφής
Κύριος συγγραφέας: Αργυριάδης, Χρήστος
Άλλοι συγγραφείς: Ζήσης, Δημήτριος
Γλώσσα:el_GR
Δημοσίευση: 2020
Θέματα:
Διαθέσιμο Online:http://hdl.handle.net/11610/20082
Ετικέτες: Προσθήκη ετικέτας
Δεν υπάρχουν, Καταχωρήστε ετικέτα πρώτοι!
Περιγραφή
Περίληψη:In this thesis, we studied the problem of automated classification of style in artistic paintings. Our aim was to design a system for the classification of paintings according to artistic style, using machine learning algorithms. From the field of machine learning we focused on artificial neural networks and we explored different network architectures and training techniques, in order to find the most suitable model for our problem. For the evaluation of the different models, we collected a database of 24.000 digitalized artistic paintings, which we used to train and evaluate our models in the task of style classification for 20 different styles. Then we included the trained network in out system, in order to allow users to classify the paintings of their choice. For the users’ interaction with the system, we designed a web application.