A Framework and Dataset for Abstract Art Generation via CalligraphyGAN

TL;DR

This paper introduces a novel AI framework combining GANs and language models to generate meaningful abstract artworks inspired by Chinese calligraphy and abstract expressionism, supported by a new dataset and user study.

Contribution

It presents a new creative framework for abstract art generation using Conditional GANs and neural language models, along with a publicly available dataset and prototype system.

Findings

Generated artworks possess intrinsic meaning and aesthetic value.

The framework outperforms existing image captioning and text-to-image methods.

User study indicates positive reception of the generated art.

Abstract

With the advancement of deep learning, artificial intelligence (AI) has made many breakthroughs in recent years and achieved superhuman performance in various tasks such as object detection, reading comprehension, and video games. Generative Modeling, such as various Generative Adversarial Networks (GAN) models, has been applied to generate paintings and music. Research in Natural Language Processing (NLP) also had a leap forward in 2018 since the release of the pre-trained contextual neural language models such as BERT and recently released GPT3. Despite the exciting AI applications aforementioned, AI is still significantly lagging behind humans in creativity, which is often considered the ultimate moonshot for AI. Our work is inspired by Chinese calligraphy, which is a unique form of visual art where the character itself is an aesthetic painting. We also draw inspirations from paintings of the Abstract Expressionist movement in the 1940s and 1950s, such as the work by American painter Franz Kline. In this paper, we present a creative framework based on Conditional Generative Adversarial Networks and Contextual Neural Language Model to generate abstract artworks that have intrinsic meaning and aesthetic value, which is different from the existing work, such as image captioning and text-to-image generation, where the texts are the descriptions of the images. In addition, we have publicly released a Chinese calligraphy image dataset and demonstrate our framework using a prototype system and a user study.