<p>Remote sensing (RS) technologies play a pivotal role in acquiring massive volumes of spatial data across various domains. To augment the capabilities of RS systems such as scene classification and change detection, machine learning (ML), especially deep learning (DL) networks, has achieved great success, but many of them often rely on a large number of labeled samples for supervision. As sufficient labeled training data are not always ready due to, e.g., continuously emerging RS datasets and costly sample annotation in real-world applications. In order to address the issue of sample scarcity, many studies prefer to utilize auxiliary information including those in the form of knowledge graph (KG) and zero-shot learning (ZSL) to reduce the reliance on labeled samples. To exclusively focus on discussing related technologies, this paper thoroughly reviews zero-shot learning (ZSL) and knowledge graphs (KG) with remote sensing (RS), highlighting their individual benefits and combined potential. Additionally, it discussed constructing methods of specific KGs for RS applications, emphasizing semantic relationships for contextualized information retrieval. The paper discusses the advantages of integrating ZSL &amp; KG methods in RS applications, particularly in scene classification, and then presents emerging approaches based on ZSLKG across various domains which quotes the potentiality of ZSLKG and finally, suggests potential RS applications complying ZSL &amp; KG. Overall, it highlights the transformative impact of ZSLKG convergence on enhancing the intelligence and efficiency of RS applications.</p>