Media Fragment Creation
Starting from a video file, media fragment creation is the process of identifying different parts of this video (i.e., fragments) that each has some meaning by itself, and therefore could be re-used independently of the rest of the video. Media fragments can be temporal, in which case we call them shots, scenes or stories, depending on how these fragments were defined and detected, or even spatio-temporal, e.g. corresponding to a specific object that appears in a video shot.
Typically, media fragment creation is achieved by applying a combination of analysis technologies to the video, which include feature extraction for video representation, feature transformation, and supervised learning as well as other machine learning techniques.