Methods and systems of spatiotemporal pattern recognition for video content development

• US 10,521,671 B2
• Filed: 05/04/2017
• Issued: 12/31/2019
• Est. Priority Date: 02/28/2014
• Status: Active Grant

• Current Assignee: Genius Sports SS LLC
• Original Assignee: Second Spectrum, Inc. (Genius Sports Ltd.)
• Inventors: Chang, Yu-Han, Maheswaran, Rajiv, Su, Jeffrey Wayne, Hollingsworth, Noel
• Primary Examiner(s): Tran, Loi H
• Application Number:
  US15/586,379
• Publication Number:
  US 20170238055A1
• Time in Patent Office: 971 Days
• Field of Search: 725 19
• CPC Class Codes: A63F 13/60 Generating or modifying game content before or while executing the game program, e.g. authoring tools specially adapted for game development or game-integrated level editor
• CPC Class Codes: G06F 3/012 Head tracking input arrangements
• CPC Class Codes: G06F 3/013 Eye tracking input arrangements G06F3/015 takes precedence
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 20/10 using kernel methods, e.g. support vector machines [SVM]
• CPC Class Codes: G06T 2207/20081 Training; Learning
• CPC Class Codes: G06T 2207/30221 Sports video; Sports image
• CPC Class Codes: G06V 20/42 of sport video content
• CPC Class Codes: G06V 20/46 Extracting features or characteristics from the video content, e.g. video fingerprints, representative shots or key frames
• CPC Class Codes: G11B 27/031 Electronic editing of digitised analogue information signals, e.g. audio or video signals
• CPC Class Codes: G11B 27/28 by using information signals recorded by the same method as the main recording G11B27/22 takes precedence
• CPC Class Codes: H04N 13/117 the virtual viewpoint locations being selected by the viewers or determined by viewer tracking
• CPC Class Codes: H04N 13/204 using stereoscopic image cameras stereoscopic photography G03B35/00
• CPC Class Codes: H04N 13/243 using three or more 2D image sensors
• CPC Class Codes: H04N 21/2187 Live feed
• CPC Class Codes: H04N 21/23418 involving operations for analysing video streams, e.g. detecting features or characteristics television picture signal circuitry for scene change detection H04N5/147; filtering for image enhancement G06T5/00; methods or arrangements for recognising scenes G06V20/00; arrangements characterised by components specially adapted for monitoring, identification or recognition of video in broadcast systems H04H60/59
• CPC Class Codes: H04N 21/251 Learning process for intelligent management, e.g. learning user preferences for recommending movies details of learning user preferences for the retrieval of video data in a video database G06F16/739; computer systems using learning methods G06N3/08
• CPC Class Codes: H04N 21/4223 Cameras H04N23/00 takes precedence
• CPC Class Codes: H04N 21/4345 Extraction or processing of SI, e.g. extracting service information from an MPEG stream
• CPC Class Codes: H04N 21/44008 involving operations for analysing video streams, e.g. detecting features or characteristics in the video stream arrangements characterised by components specially adapted for monitoring, identification or recognition of video in broadcast systems H04H60/59
• CPC Class Codes: H04N 21/4532 : involving end-user characteristics, e.g. viewer profile, preferences monitoring of user activities for profile generation for accessing a video database G06F16/739; user profiles in network data switching protocols H04L67/306; processing of user preferences or user profiles in wireless networks H04W8/18
• CPC Class Codes: H04N 21/4662 : characterized by learning algorithms
• CPC Class Codes: H04N 21/8456 : by decomposing the content in the time domain, e.g. in time segments
• CPC Class Codes: H04N 21/8549 : Creating video summaries, e.g. movie trailer retrieval in video databases by using presentations in form of a video summary G06F16/739
• CPC Class Codes: H04N 5/2224 : related to virtual studio applications

Abstract

Providing enhanced video content includes processing at least one video feed through at least one spatiotemporal pattern recognition algorithm that uses machine learning to develop an understanding of a plurality of events and to determine at least one event type for each of the plurality of events. The event type includes an entry in a relationship library detailing a relationship between two visible features. Extracting and indexing a plurality of video cuts from the video feed is performed based on the at least one event type determined by the understanding that corresponds to an event in the plurality of events detectable in the video cuts. Lastly, automatically and under computer control, an enhanced video content data structure is generated using the extracted plurality of video cuts based on the indexing of the extracted plurality of video cuts.

Claims

1. A method for providing enhanced video content, comprising:

processing at least one video feed through at least one spatiotemporal pattern recognition algorithm that uses machine learning to determine at least one event type for each of a plurality of events within the at least one video feed, wherein machine learning determines the at least one event type for at least one spatiotemporal pattern selected from the group consisting of relative motion of two visible features toward each other for at least a duration threshold, acceleration of motion of at least two visible features with respect to each other being greater than an acceleration threshold, rate of motion of two visible features toward each other, projected point of intersection of the two visible features, and separation distance between the two visible features being less than a separation threshold;

extracting a plurality of video cuts from the at least one video feed;

indexing the extracted plurality of video cuts based on the at least one event type determined by the machine learning that corresponds to an event in the plurality of events detectable in the plurality of video cuts; and

automatically, under computer control, generating an enhanced video content data structure using the extracted plurality of video cuts based on the indexing of the extracted plurality of video cuts,

wherein the at least one spatiotemporal pattern recognition algorithm is based on at least one pattern recognized by adjusting an input feature and a weight within a machine learning system, wherein the input feature is selected from the group consisting of relative direction of motion of at least two visible features, duration of relative motion of visible features with respect to each other, rate of motion of at least two visible features with respect to each other, acceleration of motion of at least two visible features with respect to each other, projected point of intersection of at least two visible features with respect to each other and separation distance between at least two visible features with respect to each other; and

wherein extracting the plurality of video cuts includes automatically extracting a cut from the video feed based on a result of processing another input feed with the machine learning, the another input feed including at least one of a portion of content of a broadcast commentary and a change in camera view in the another input feed.

2. The method of claim 1, wherein the at least one spatiotemporal pattern recognition algorithm is based on at least one pattern recognized by adjusting at least one of an input feature and a weight within a machine learning system, wherein the input feature is selected from the group consisting of relative direction of motion of at least two visible features, duration of relative motion of visible features with respect to each other, rate of motion of at least two visible features with respect to each other, acceleration of motion of at least two visible feature with respect to each other, projected point of intersection of at least two visible features with respect to each other and separation distance between at least two visible features with respect to each other.

3. The method of claim 1, wherein generating the enhanced video content data structure is based at least in part on at least one of a user preference and a user profile for a user for which the enhanced video content data structure is generated.

4. The method of claim 1, further comprising providing a user interface for display on a mobile device, wherein the user interface includes at least one of a search option and a filtering option to allow a user to at least one of specify and select a description of a type of event, wherein the enhanced video content data structure is generated to match the description.

5. The method of claim 1, wherein using the machine learning further comprises using the plurality of events in position tracking data over time obtained from at least one of the at least one video feed and a chip-based player tracking system, and wherein the machine learning is based on at least two of spatial configuration, relative motion, and projected motion of at least one of a player and an item used in a game.

6. The method of claim 1, wherein using the machine learning further comprises aligning multiple unsynchronized input feeds related to an event of the plurality of events using at least one of a hierarchy of algorithms and a hierarchy of human operators, wherein the unsynchronized input feeds are selected from the group consisting of one or more broadcast video feeds of the event, one or more feeds of tracking video for the event, and one or more play-by-play data feeds of the event.

7. The method of claim 6, wherein the multiple unsynchronized input feeds include at least three feeds selected from at least two event types related to the event.

8. The method of claim 6, further comprising at least one of validating and modifying the alignment of the unsynchronized input feeds using a hierarchy involving at least two of one or more algorithms, one or more human operators, and one or more input feeds.

9. The method of claim 1, further comprising at least one of validating and modifying the machine learning using a hierarchy involving at least two of one or more algorithms, one or more human operators, and one or more input feeds.

10. The method of claim 1, further comprising automatically developing a semantic index of the at least one video feed based on an output of the machine learning of at least one event of the plurality of events in the video feed to indicate a game time of the at least one event in the video feed and a location of a display of the at least one event in the video feed.

11. The method of claim 10, wherein the location of the display of the at least one event in the video feed includes at least one of a pixel location, a voxel location, a raster image location.

12. The method of claim 10, further comprising providing the semantic index of the video feed with the video feed configured to enable semantic-based augmentation of the video feed.

13. The method of claim 12, wherein augmentation of the video feed includes adding content based on the identified location and enabling at least one of a touch interface feature and a mouse interface feature based on the identified location.

14. A method, comprising:

processing at least one video feed through at least one spatiotemporal pattern recognition algorithm that uses machine learning to determine at least one event type for each of a plurality of events within the at least one video feed, wherein machine learning determines the at least one event type for at least one spatiotemporal pattern selected from the group consisting of relative motion of two visible features toward each other for at least a duration threshold, acceleration of motion of at least two visible features with respect to each other being greater than an acceleration threshold, rate of motion of two visible features toward each other, projected point of intersection of the two visible features, and separation distance between the two visible features being less than a separation threshold;

extracting a plurality of video cuts from the at least one video feed;

indexing the plurality of video cuts based on the at least one event type determined by machine learning; and

providing a mobile application having a user interface configured to permit a user to find the extracted plurality of video cuts based on the indexing of the extracted plurality of video cuts with the mobile application,

wherein the at least one spatiotemporal pattern recognition algorithm is based on at least one pattern recognized by adjusting an input feature and a weight within a machine learning system, wherein the input feature is selected from the group consisting of relative direction of motion of at least two visible features, duration of relative motion of visible features with respect to each other, rate of motion of at least two visible features with respect to each other, acceleration of motion of at least two visible features with respect to each other, projected point of intersection of at least two visible features with respect to each other and separation distance between at least two visible features with respect to each other; and

wherein extracting the plurality of video cuts includes automatically extracting a cut from the video feed based on a result of processing another input feed with the machine learning, the another input feed including at least one of a portion of content of a broadcast commentary and a change in camera view in the another input feed.

15. The method of claim 14, the at least one spatiotemporal pattern recognition algorithm is based on at least one pattern recognized by adjusting at least one of an input feature and a weight within a machine learning system, wherein the input feature is selected from the group consisting of relative direction of motion of at least two visible features, duration of relative motion of visible features with respect to each other, rate of motion of at least two visible features with respect to each other, acceleration of motion of at least two visible feature with respect to each other, projected point of intersection of at least two visible features with respect to each other and separation distance between at least two visible features with respect to each other.

16. The method of claim 14, including generating at least one metric associated with at least one event of the plurality of events and wherein the user interface of the mobile application is configured to permit the user to enhance a video cut from the plurality of video cuts by selecting the metric to be included in the video cut.

17. The method of claim 14, wherein the user interface of the mobile application is configured to permit the user to share an edited video via the mobile application.

18. The method of claim 14, wherein using the machine learning further comprises using the plurality of events in position tracking data over time obtained from at least one of the at least one video feed and a chip-based player tracking system, and wherein the machine learning is based on at least two of spatial configuration, relative motion, and projected motion of at least one of a player and an item used in a game.

19. The method of claim 14, wherein using the machine learning further comprises aligning multiple unsynchronized input feeds related to an event of the plurality of events using at least one of a hierarchy of algorithms and a hierarchy of human operators, wherein the unsynchronized input feeds are selected from the group consisting of one or more broadcast video feeds of the event, one or more feeds of tracking video for the event, and one or more play-by-play data feeds of the event.

20. The method of claim 19, wherein the multiple unsynchronized input feeds include at least three feeds selected from at least two types related to the event of the plurality of events.

21. The method of claim 19, further comprising at least one of validating and modifying the alignment of the unsynchronized input feeds using a hierarchy involving at least two of one or more algorithms, one or more human operators and one or more input feeds where at least one algorithm in the hierarchy for validation is based on a nature of the input feed.

22. The method of claim 14, further comprising at least one of validating the determination and modifying the determination using a hierarchy involving at least two of one or more algorithms, one or more human operators and one or more input feeds where at least one algorithm in the hierarchy for validating is based on a nature of the input feed.