Cervical cancer is one of the most common malignant tumors in women; hence the world has been working to improve the effective screening and prevention of cervical cancer. Colposcopy plays a central role in cervical cancer prevention, but its accuracy and reproducibility are still limited. The use of deep learning in the field of medical images allows more researchers as well as to explore the application of deep learning in colposcopy image-assisted diagnosis. In this paper, we summarize the research status of this field and propose the current shortcomings and improvement directions this research field.