GSGNet-S*: Graph Semantic Guidance Network via Knowledge Distillation for Optical Remote Sensing Image Scene Analysis

In recent years, optical remote sensing image (ORSI) scene analysis has attracted increasing interest. However, existing networks show a trend of bifurcation. Lightweight networks have very high inference speed but poor inference of contextual information in highly complex backgrounds. In contrast, networks with high-performance contextual information reasoning capability require many parameters and are computationally expensive. Since the knowledge distillation (KD) method can greatly lighten the model, we propose a graph semantic guided network (GSGNet) that utilizes knowledge refinement for ORSI scenario analysis, which has a high inference speed while maintaining practical contextual inference capability. Rich semantic and detailed information facilitates the semantic segmentation of ORSIs. We design adjacent dynamic capture (ADC) and local–global map inference modules that can effectively extract low-level spatial details and high-level contextual semantics. To improve the attention map relearning (AR) performance of the distillation method, we designed semantically guided fusion modules (SGFMs) to locate spatial information and refine edge information. We also employed a structural relationship transfer (SRT) distillation method in which the structural relationship knowledge of the teacher model (GSGNet-T) was used to guide the student model (GSGNet-S). We compared the performances of GSGNet-T and the GSGNet-S with KD (GSGNet-S*) with those of several state-of-the-art (SOAT) methods on the Vaihingen and Potsdam datasets. Extensive experiments showed that GSGNet-S* outperformed most advanced methods with only 19.61 M parameters and a computation cost of 2.9 GFLOPs. The experimental results and code of our network can be accessed at the following URL:
https://github.com/LYZ00918/GSGNet-KD
.