Multi-Task Semantic Communication for Remote Sensing
Qing Bian, Zhicheng Bao, Dong Chen, Xiaodong Xu, Yunfei Luo, Rui Meng · 2025
This paper proposes a multi-task remote sensing semantic communication framework (RSSC). The framework performs image encoding and applies a gate net selective mechanism at the transmitter, while simultaneously executing image reconstruction and object detection at the receiver. The gate net module can be enabled or disabled based on choice. The role of the gate net is to filter out images and allow only those that meet the receiver’s requirements to pass through. Additionally, by employing distillation learning, the gate net effectively enhances performance. Experimental results demonstrate that the proposed method exhibits satisfactory performance in both image reconstruction and object detection, especially under low signal-to-noise ratio (SNR) conditions. The gate net is capable of efficiently controlling images under low computing power conditions.