Content-based video segmentation

摘要

In general, video segmentation techniques are described. According to various examples, the video segmentation techniques may be based on video content. An example method includes determining one or more segments into which to divide video content, dividing the video content into the determined number of segments identifying a boundary frame associated with each of the segments, and adjusting the respective boundary frame associated with a first segment of the segments to generate an adjusted boundary frame associated with the first segment, wherein the adjusting is based on an one or more entity representations associated with the adjusted boundary frame.

主权项

1. A method comprising:
determining, by a computing device, one or more segments into which to divide video content; dividing, by the computing device, the video content into the one or more segments; identifying, by the computing device, a respective boundary frame associated with each of the segments; identifying, by the computing device, a frame with a greatest activity level among a set of frames based on one or more entity representations associated with the set of frames, wherein each frame of the set of frames occurs within a predetermined time span from the respective boundary frame associated with a first segment of the one or more segments; and adjusting, by the computing device, the respective boundary frame associated with the first segment to generate an adjusted boundary frame associated with the first segment at least in part by identifying the frame with the greatest activity level as the adjusted boundary frame.