Scopus Indexed Publications

Paper Details


Title
Genre Recognition of Artworks using Convolutional Neural Network
Author
Md. Kamran Hosain, Harun-Ur-Rashid,
Email
Abstract
Artists always try to express their feelings by creating aesthetic images. These artistic items covered with many untold sensations and unpublished articulation. According to the objects and depicted themes, artworks can be divided by genre. The genre is a procedure to learn about arts by identifying the similarities and to point out groups of dissimilar works of arts as well. It also helps to know about the aesthetic characteristics. Consequently, the goal of this research mainly focused to predict genre of the artworks. A state-of-the-art deep learning method, Convolutional Neural Networks (CNN) is used for the prediction tasks. The image classification experiment is executed with a variation in typical CNN architecture along with two other models- VGG-16 and InceptionV3 for multi-label dataset. The modified CNN model gives a satisfactory result to predict the genre of a particular artwork with an accuracy of 98.21%.

Keywords
Artworks , Genre prediction , Deep learning , Convolutional neural network
Journal or Conference Name
23rd International Conference on Computer and Information Technology (ICCIT)
Publication Year
2020
Indexing
scopus