Measuring Similarity between Artistic and AI Generated Images using Siamese Neural Networks

A new arXiv study (2608.28671v1) quantifies similarity between original artworks and images generated by Stable Diffusion XL Refiner 1.0, reporting that prior studies found up to 81% style replication and 90% visual similarity. Using Siamese Networks with frozen CLIP encoders and cosine similarity optimized through triplet loss, the researchers achieved 99.9% training accuracy and 99.4% test accuracy with strong inter-class separation (δμ = 0.677), demonstrating effective semantic-visual embeddings for detecting AI-generated art plagiarism.

arXiv:2608.28671v1 Announce Type: new Abstract: AI-generated art has sparked debates around potential plagiarism, as these images may closely resemble existing artworks. This research quantifies the similarity between original pieces and AI-generated counterparts, particularly those produced by the Stable Diffusion XL Refiner 1.0. We use Siamese Networks with frozen CLIP encoders and cosine similarity optimized through triplet loss. A dataset of paired original and generated images was built using image-to-image generation and custom prompts, enriched with semantic descriptors and BLIP-2 captions. Prior studies report up to 81\% style replication and 90\% visual similarity. Our results show high discriminative performance: training accuracy reached 99.9\%, and the best model configuration achieved 99.4\% test accuracy with strong inter-class separation $\delta \mu$ = 0.677 , demonstrating the effectiveness of our semantic-visual embeddings.