US 12,294,755 B2
Livestream key moment identification
Sunav Choudhary, West Bengal (IN); Atanu R. Sinha, Bangalore (IN); Sarthak Chakraborty, Kolkata (IN); Sai Shashank Kalakonda, Telangana (IN); Liza Dahiya, Haryana (IN); Purnima Grover, Haryana (IN); and Kartavya Jain, Rajasthan (IN)
Assigned to ADOBE INC., San Jose, CA (US)
Filed by ADOBE INC., San Jose, CA (US)
Filed on Feb. 28, 2023, as Appl. No. 18/176,114.
Prior Publication US 2024/0292046 A1, Aug. 29, 2024
Int. Cl. H04N 21/262 (2011.01); H04N 21/2187 (2011.01); H04N 21/233 (2011.01); H04N 21/234 (2011.01); H04N 21/25 (2011.01); H04N 21/442 (2011.01); H04N 21/4788 (2011.01); H04N 21/81 (2011.01)
CPC H04N 21/262 (2013.01) [H04N 21/2187 (2013.01); H04N 21/233 (2013.01); H04N 21/23418 (2013.01); H04N 21/251 (2013.01); H04N 21/44218 (2013.01); H04N 21/4788 (2013.01); H04N 21/812 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A method comprising:
obtaining video data and chat log text, wherein the chat log text is aligned with a timeline of the video data;
encoding the video data and the chat log text using a fusion encoder model to obtain a combined feature vector representing the video data and the chat log text;
generating a moment importance score for a time of the video data by decoding the combined feature vector using a decoder of a machine learning model, wherein the moment importance score indicates a probability that the time of the video comprises a key moment; and
presenting content to a user at the time of the video data based on the moment importance score.