The invention discloses a face living body detection method based on a video, wherein an off-line training module and an on-line detection module are adopted in the face living body detection method.The operating process of the off-line training module comprises the steps of: S11. acquiring a large number of face video data through a camera; S12. pre-processing the acquired video data; and S13. training the pre-processed video data in the form of deep learning. The operating process of the on-line detection module comprises the steps of: S21. acquiring video data of a to-be-detected object through a camera; S22. pre-processing the acquired video data of the to-be-detected object; and S23. performing living body detection on the to-be-detected object. A detection result is compared with apreset threshold to obtain a prediction result, so as to judge whether the detected object is a living body. In comparison with the prior art, the face living body detection method of the invention not only uses information of a single picture, but also uses information among various frames of pictures in the video, thereby improving accuracy rate of face living body detection. In addition, the face living body detection method improves precision of face living body detection, and improves security of the fields such as finance and security related to the living body.