US 11,758,233 B1
Time marking chapters in media items at a platform using machine-learning
Chenjie Gu, Mountan View, CA (US); Wei-Hong Chuang, Mountain View, CA (US); Min-Hsuan Tsai, San Jose, CA (US); Jianfeng Yang, Mountain View, CA (US); Ji Zhang, Mountain View, CA (US); Honglu Zhou, Highland Park, NJ (US); and Hassan Akbari, New York, NY (US)
Assigned to Google LLC, Mountain View, CA (US)
Filed by Google LLC, Mountain View, CA (US)
Filed on Jun. 8, 2022, as Appl. No. 17/835,547.
Int. Cl. H04N 21/472 (2011.01); G11B 27/34 (2006.01)
CPC H04N 21/47217 (2013.01) [G11B 27/34 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A method comprising:
identifying a media item to be provided to one or more users of a platform;
determining that a duration of the identified media item exceeds a span of an input window associated with a machine-learning model;
providing an indication of the identified media item as input to the machine-learning model, wherein the machine-learning model is trained using different feature types of historical media items to predict, for a given media item and based on the input window, a plurality of content segments of the given media item each depicting, to the one or more users, a distinct section of the media item;
obtaining a first set of outputs of the machine-learning model, wherein the first set of obtained outputs comprise time marks identifying a first set of content segments for an initial duration of the media item;
applying the input window to a subsequent duration of the media item that follows the initial duration of the media item;
obtaining a second set of outputs of the machine-learning model, wherein the second set of obtained outputs comprise time marks identifying a second set of content segments for the subsequent duration of the media item;
associating each content segment in the first set of content segments and the second set of content segments with a segment start indicator for a timeline of the media item; and
providing the media item and an indication of each segment start indicator for presentation to at least one user of the one or more users.