The invention discloses a shot division method and system based on video analysis. The method includes the following steps: reading and successively obtaining all video frames included in video fragments; on the basis of continuity of the adjacent video frames, dividing the video fragments and obtaining all lens information forming the video fragments; carrying out frame sampling on the video frames in each piece of lens information, obtaining corresponding sampling points, matching the sampling points with a human body sample feature database, and obtaining matched sampling point values; when the matched sampling point values are smaller than a pre-defined matched threshold and the lens information does not include a figure frame, judging that the shot of a lens is a long shot; when the matched sampling point values are greater than or equal to the pre-defined matched threshold and the lens information includes the figure frame, obtaining a ratio of the area of the head of a figure in the lens information to the area of a display screen; and dividing shots of the lens. The shot division method and system help to solve the problems about high cost and low efficiency caused by artificial subjective shot identification.