In this paper, the relevant literature on cross-modal retrieval in WOS database is used. Combining bibliometrics and visual knowledge mapping analysis methods, Hiscite and CiteSpace analysis tools are used to conduct multi-dimensional statistical analysis and visual analysis of selected documents. Through the research on the main authors, countries, institutions, hotspots, research trends and research frontiers in the field, this paper finds that China is a country with more research results on cross-modal retrieval and research direction is mainly focused on algorithm improvement, model research,etc. The research frontiers are mainly concentrated on artificial intelligence,deep learning and multimedia recognition.