Derived competition
Video feeds may be analyzed to identify matchups between participants. The matchups may be used to derive a competition, context, statistics, and the like. A result of the derived competition may depend, at least in part, on an ordering and outcomes of a set of matchup events. The derived competition may be presented as a collection of video clips corresponding to the set of matchup events, related commentary, and/or related statistics.
1. A method comprising:
accessing a set of stored video feeds;
receiving first rules for detecting matchups, wherein the first rules include spatiotemporal pattern algorithms that use a machine learning system to facilitate detecting a set of matchup events in the set of stored video feeds;
receiving second rules for detecting participants in the set of stored video feeds;
processing, with the first rules and the second rules, frames of the set of stored video feeds to identify at least one matchup event and at least one participant associated with each identified matchup event;
determining, in the identified at least one matchup event, event features;
tabulating the determined event features,
receiving weighting for the participants; and
determining ranking of the set of matchup events based on the weighting and the tabulation of the determined event features.
2. The method of claim 1 , wherein tabulating further comprises determining a number of matchup events that includes a first participant and determining outcomes of the matchup events that include the first participant.
3. The method of claim 1 , further comprising generating, based on the tabulated event features, a recommendation for monitoring video feeds for matchup events that include a first participant.
4. The method of claim 3 , further comprising, receiving a selection for monitoring and analyzing a second set of video feeds for videos relating to one or more matchups that include the first participant.
5. The method of claim 4 , further comprising:
generating an alert for a user in response to detecting the one or more matchups that include the first participant.
6. The method of claim 4 , further comprising:
generating a video summary of the one or more matchups that include the first participant by extracting portions of video content from the set of stored video feeds.
7. The method of claim 1 , wherein the first rules include a trained machine learning model.
8. The method of claim 1 , wherein determining event features comprises determining at least one of: a type of event, a duration of the event, an outcome of the event, or a crowd reaction to the event.
9. The method of claim 1 , further comprising:
determining categories of matchup events based on the tabulation.
10. The method of claim 1 , wherein the second rules include rules for jersey number detection.
11. The method of claim 10 , wherein the jersey number detection includes video frame by frame jersey number detection comprising classifying a plurality of windows in each frame as one of containing a number and as numberless.
12. The method of claim 11 , wherein the classified plurality of windows in each frame containing the number are stitched together to form a temporal jersey number track.