Advancing Image Transfer Through Semantic-Aided Approaches: A Multi-Modal Exploration

Semantic Communication (SC), in contrast to conventional communication, prioritizes meaning over raw data, thereby minimizing errors. In this paper, we advance image transfer using semantic-aided approaches. Leveraging deep learning models such as Bootstrapping Language-Image Pre-Training and Text-to-Image Diffusion Models, we initially utilize captions as a single mode to reduce data transfer size and convey image content. However, recognizing the structural importance of images, we introduce a second mode, favoring line art for its efficacy in depicting image structure. Our findings highlight the potential of multi-modal approaches to improving SC systems for various applications in the 6G era.