Multimodal knowledge graph is an emerging research direction in the field of knowledge graph, which integrates different modal data (such as text, image, audio, etc.) into a unified knowledge graph to more comprehensively describe and understand the real world. This article reviews the current research progress and future directions of multimodal knowledge graphs. First, the article introduces the basic concepts and importance of multimodal knowledge graphs, emphasizing its application potential in artificial intelligence and data science. Next, the construction methods of multi-modal knowledge graphs are reviewed, including the integration of multi-modal data, feature extraction and cross-modal association learning. In addition, the article also discusses the specific applications of multi-modal knowledge graphs in natural language processing, computer vision, healthcare and other fields to demonstrate its value in multiple fields. Finally, this article proposes the future development direction of multimodal knowledge graph research, including more effective data fusion technology, application of deep learning models, research on ethics and privacy issues, etc. Multimodal knowledge graph is a research field full of potential that will continue to promote the development of artificial intelligence and data science and bring innovation and opportunities to applications in multiple fields.