🤖 AI Summary
To address the challenges of detail loss in photographic content, structural distortion, and insufficient ink-brush texture in unpaired image style transfer to Chinese ink painting, this paper proposes a GAN-based framework integrated with saliency detection. Our method balances global composition and local texture fidelity without requiring paired training data. Key contributions include: (1) an SIOU loss that explicitly enforces structural consistency; (2) saliency-adaptive normalization (SANorm), which dynamically modulates feature normalization strength according to salient regions; and (3) a saliency-guided discriminator that enhances discrimination capability over semantically critical areas. Quantitative and qualitative evaluations demonstrate that our approach outperforms state-of-the-art GANs and diffusion models in FID, LPIPS, and human assessment. It achieves significant improvements in content fidelity, brushstroke expressiveness, and natural ink diffusion—hallmarks of authentic Chinese ink aesthetics.
📝 Abstract
Recent style transfer problems are still largely dominated by Generative Adversarial Network (GAN) from the perspective of cross-domain image-to-image (I2I) translation, where the pivotal issue is to learn and transfer target-domain style patterns onto source-domain content images. This paper handles the problem of translating real pictures into traditional Chinese ink-wash paintings, i.e., Chinese ink-wash painting style transfer. Though a wide range of I2I models tackle this problem, a notable challenge is that the content details of the source image could be easily erased or corrupted due to the transfer of ink-wash style elements. To remedy this issue, we propose to incorporate saliency detection into the unpaired I2I framework to regularize image content, where the detected saliency map is utilized from two aspects: (
omannumeral1) we propose saliency IOU (SIOU) loss to explicitly regularize object content structure by enforcing saliency consistency before and after image stylization; (
omannumeral2) we propose saliency adaptive normalization (SANorm) which implicitly enhances object structure integrity of the generated paintings by dynamically injecting image saliency information into the generator to guide stylization process. Besides, we also propose saliency attended discriminator which harnesses image saliency information to focus generative adversarial attention onto the drawn objects, contributing to generating more vivid and delicate brush strokes and ink-wash textures. Extensive qualitative and quantitative experiments demonstrate superiority of our approach over related advanced image stylization methods in both GAN and diffusion model paradigms.