CPC A63F 13/86 (2014.09) [A63F 13/537 (2014.09); G06F 18/217 (2023.01); G06F 18/22 (2023.01); G06V 20/41 (2022.01); G06V 20/46 (2022.01); H04N 19/42 (2014.11)] | 16 Claims |
1. A computer-implemented method for extracting information from videos, comprising:
receiving a set of annotated screen captures comprising portions of interest within corresponding screen captures of a video game;
receiving at least a first set of videos for the video game that comprise gameplay of the video game generated through interactions of one or more players with the video game;
training a first machine-learning model to identify the portions of interest within the first set of videos based on the set of annotated screen captures;
generating validation data based on results of the first machine-learning model;
receiving a second set of videos for the video game that comprise gameplay of the video game generated by a variety of players, the second set of videos being larger than the first set of videos;
training a second machine-learning model to identify the portions of interest within the second set of videos based on the validation data;
extracting information based on the portions of interest identified in at least the first set of videos and the second set of videos; and
extracting information based on the portions of interest identified in at least the first set of videos.
|