Generative Semantic Coding for Ultra-Low Bitrate Visual Communication and Analysis
This addresses the problem of enabling reliable visual communication and analysis in extreme low-bandwidth environments like deep space or battlefields, representing an incremental improvement over existing methods.
The paper tackles ultra-low bitrate visual communication for remote analysis and human interactions in bandwidth-limited scenarios, proposing a method that integrates image generation with deep compression to achieve comparable reconstruction quality and analysis accuracy with significantly reduced bandwidth.
We consider the problem of ultra-low bit rate visual communication for remote vision analysis, human interactions and control in challenging scenarios with very low communication bandwidth, such as deep space exploration, battlefield intelligence, and robot navigation in complex environments. In this paper, we ask the following important question: can we accurately reconstruct the visual scene using only a very small portion of the bit rate in existing coding methods while not sacrificing the accuracy of vision analysis and performance of human interactions? Existing text-to-image generation models offer a new approach for ultra-low bitrate image description. However, they can only achieve a semantic-level approximation of the visual scene, which is far insufficient for the purpose of visual communication and remote vision analysis and human interactions. To address this important issue, we propose to seamlessly integrate image generation with deep image compression, using joint text and coding latent to guide the rectified flow models for precise generation of the visual scene. The semantic text description and coding latent are both encoded and transmitted to the decoder at a very small bit rate. Experimental results demonstrate that our method can achieve the same image reconstruction quality and vision analysis accuracy as existing methods while using much less bandwidth. The code will be released upon paper acceptance.