Video segmentation method and system

The invention provides a video segmentation method, which comprises the following steps of carrying out scene segmentation on a target video stream, recognizing whether a facial image and/or features of five sense organs of a face exist in each video scene and obtaining a face picture set included in the video scene and a clothing picture set corresponding to the face picture set; and respectively carrying out clustering analysis on the face picture set and the clothing picture set, when an obtained clustering analysis result is smaller than a preset threshold, judging the video scene as a break point of the target video stream and carrying out segmentation on the target video stream according to each video scene which is taken as the break point. The method directly judges whether each scene after segmentation of the target video is the break point of the target video and takes the video scene which meets the condition as the break point to carry out segmentation on the target video stream without establishi