PROJECT TITLE :

A Deep Learning Perspective on Beauty Sentiment and Remembrance of Art

ABSTRACT:

New research opportunities in the confluence of artificial intelligence and art are opening up as massive digitized fine art collections emerge and deep learning approaches perform well. We use convolutional neural networks (CNN) to predict scores related to three subjective aspects of human perception: aesthetic evaluation of the image, sentiment evoked by the image, and memorability of the image, in order to investigate the applicability of deep learning techniques in understanding art images beyond object recognition and classification. We assess multiple distinct CNN models trained on diverse natural image datasets for each topic and choose the best performing model based on qualitative results and comparisons with current subjective artwork judgments. In addition, we use a variety of decision tree-based machine learning models to assess the relative importance of various image attributes such as content, composition, and color in calculating image aesthetics, visual sentiment, and memorability scores. Our findings imply that aesthetics are influenced by both content and image illumination, with color vividness and harmony having a large influence on sentiment prediction and object emphasis having a strong impact on memorability. In addition, we examine the distribution of anticipated aesthetic, sentiment, and memorability scores in the context of art history, looking at different artistic styles, genres, artists, and centuries. The approach given here allows for new ways of examining fine art collections based on highly subjective qualities of art, as well as bridging the gap between traditional formal analysis and computational fine art analysis.