Can AI Recognize the Style of Art? Analyzing Aesthetics through the Lens of Style Transfer
Yunha Yeo, Daeho Um
TL;DR
The paper questions whether AI can recognize and transfer artistic style in a way aligned with human perception. It compares CNN-based and Transformer-based style transfer to identify which style elements they capture, including color, texture, global composition, signs, and culture, and analyzes their limitations from an aesthetic perspective. Results show CNN favors surface-level attributes while Transformers grasp global context, yet neither fully captures the deeper artistic intentions or mental imagery behind styles like Cubism or ancient art. The authors advocate future research that emphasizes aesthetic value and perceptual authenticity beyond mere surface replication, with implications for AI-driven art and computational aesthetics.
Abstract
This study investigates how artificial intelligence (AI) recognizes style through style transfer-an AI technique that generates a new image by applying the style of one image to another. Despite the considerable interest that style transfer has garnered among researchers, most efforts have focused on enhancing the quality of output images through advanced AI algorithms. In this paper, we approach style transfer from an aesthetic perspective, thereby bridging AI techniques and aesthetics. We analyze two style transfer algorithms: one based on convolutional neural networks (CNNs) and the other utilizing recent Transformer models. By comparing the images produced by each, we explore the elements that constitute the style of artworks through an aesthetic analysis of the style transfer results. We then elucidate the limitations of current style transfer techniques. Based on these limitations, we propose potential directions for future research on style transfer techniques.