Semantic Communication based on Large Language Model for Underwater Image Transmission

• DOI: 10.48550/arxiv.2408.12616
• Publication Date: 2024-08-08
• Authors: Weilong Chen; Wenxuan Xu; Haoran Chen; Xinran Zhang; Zhijin Qin; Yanru Zhang; Han Zhu
• Keywords: FOS: Computer and information sciences; Artificial Intelligence (cs.AI); Computer Science - Artificial Intelligence; Computer Vision and Pattern Recognition (cs.CV); Computer Science - Computer Vision and Pattern Recognition

Underwater communication is essential for environmental monitoring, marine biology research, and underwater exploration. Traditional faces limitations like low bandwidth, high latency, susceptibility to noise, while semantic (SC) offers a promising solution by focusing on the exchange of semantics rather than symbols or bits. However, SC encounters challenges in environments, including information mismatch difficulties accurately identifying transmitting critical that aligns with diverse requirements applications. To address these challenges, we propose novel Semantic Communication framework based Large Language Models (LLMs). Our leverages visual LLMs perform compression prioritization image data according query from users. By encoding key elements within images, system selectively transmits high-priority applying higher rates less regions. On receiver side, an LLM-based recovery mechanism, along Global Vision ControlNet Key Region networks, aids reconstructing thereby enhancing efficiency robustness. reduces overall size 0.8\% original. Experimental results demonstrate our method significantly outperforms existing approaches, ensuring high-quality, semantically accurate reconstruction.