Generating a video segment of an action from a video

1. A system comprising:

one or more processors; and

a memory coupled to the one or more processors, with instructions stored thereon that, when executed by the one or more processors, cause the one or more processors to perform operations comprising:

receiving a video that includes multiple frames;

identifying video segments from the video that each include a corresponding action, wherein each video segment of the video segments include two or more contiguous frames of the multiple frames of the video;

generating a confidence score for each of the video segments based on a probability that the corresponding action in the video segment corresponds to a particular action in a set of predetermined actions; and

selecting a subset of the video segments based on the confidence score for each of the video segments.

2. The system of claim 1 , wherein the operations further comprise:

generating a user profile for a user that includes a list of types of videos that the user previously indicated a preference for; and

determining a type for the video, wherein the confidence score is based on the probability that the corresponding action in the video segment corresponds to the type of video in the set of predetermined actions.

3. The system of claim 1 , wherein the operations further comprise:

identifying one or more people in the video segments;

determining that a user has a relationship with the one or more people in a first video segment; and

assigning a higher segment score to the first video segment than one or more other video segments that do not depict a person that is related to the user.

4. The system of claim 1 , wherein the video is a live video and the set of predetermined actions includes at least one of laughing or crying.

5. The system of claim 1 , wherein the operations further include identifying the action in the video based on using an audio classifier that analyzes an audio portion of the video to identify a start time and an end time of the action and to identify a type of action that is associated with audio.

6. The system of claim 1 , wherein the operations further include identifying the action in the video based on using an RGB classifier that employs a feature vector and clustering techniques to identify features that form a pattern associated with the action and a type of action.

7. The system of claim 1 , wherein the operations further include identifying the action in the video based on using a motion classifier that identifies an optical flow in the video that is associated with particular kinds of movement.

8. The system of claim 1 , wherein selecting the subset of the video segments based on the confidence score includes selecting the subset of the video segments that exceed a confidence score threshold and the operations further comprise:

generating a video clip that includes the subset of the video segments.

9. The system of claim 1 , wherein the operations further comprise:

generating a segment score to each of the video segments based on personalization information for a user, wherein selecting the subset of the video segments is further based on a combination of the confidence score and the segment score for each of the video segments exceeding a threshold animation score; and

generating a video clip that includes the subset of the video segments.

10. A method comprising:

receiving a video that includes multiple frames;

identifying video segments from the video that each include a corresponding action, wherein each video segment of the video segments include two or more contiguous frames of the multiple frames of the video;

generating a confidence score for each of the video segments based on a probability that the corresponding action in the video segment corresponds to a particular action in a set of predetermined actions; and

selecting a subset of the video segments based on the confidence score for each of the video segments.

11. The method of claim 10 , further comprising:

generating a user profile for a user that includes a list of types of videos that the user previously indicated a preference for; and

determining a type for the video, wherein the confidence score is based on the probability that the corresponding action in the video segment corresponds to the type of video in the set of predetermined actions.

12. The method of claim 10 , further comprising:

identifying one or more people in the video segments;

determining that a user has a relationship with the one or more people in a first video segment; and

assigning a higher segment score to the first video segment than one or more other video segments that do not depict a person that is related to the user.

13. The method of claim 10 , further comprising identifying the action in the video based on using an audio classifier that analyzes an audio portion of the video to identify a start time and an end time of the action and to identify a type of action that is associated with audio.

14. The method of claim 10 , further comprising identifying the action in the video based on using an RGB classifier that employs a feature vector and clustering techniques to identify features that form a pattern associated with the action and a type of action.

15. The method of claim 10 , further comprising identifying the action in the video based on using a motion classifier that identifies an optical flow in the video that is associated with particular kinds of movement.

16. A non-transitory computer readable medium with instructions that, when executed by one or more computers, cause the one or more computers to perform operations, the operations comprising:

receiving a video that includes multiple frames;

identifying video segments from the video that each include a corresponding action, wherein each video segment of the video segments include two or more contiguous frames of the multiple frames of the video;

generating a confidence score for each of the video segments based on a probability that the corresponding action in the video segment corresponds to a particular action in a set of predetermined actions; and

selecting a subset of the video segments based on the confidence score for each of the video segments.

17. The non-transitory computer readable medium of claim 16 , wherein the operations further comprise:

generating a user profile for a user that includes a list of types of videos that the user previously indicated a preference for; and

determining a type for the video, wherein the confidence score is based on the probability that the corresponding action in the video segment corresponds to the type of video in the set of predetermined actions.

18. The non-transitory computer readable medium of claim 16 , wherein the operations further comprise:

identifying one or more people in the video segments;

determining that a user has a relationship with the one or more people in a first video segment; and

assigning a higher segment score to the first video segment than one or more other video segments that do not depict a person that is related to the user.

19. The non-transitory computer readable medium of claim 16 , wherein the operations further comprise identifying the action in the video based on using an audio classifier that analyzes an audio portion of the video to identify a start time and an end time of the action and to identify a type of action that is associated with audio.

20. The non-transitory computer readable medium of claim 16 , wherein the operations further include identifying the action in the video based on using an RGB classifier that employs a feature vector and clustering techniques to identify features that form a pattern associated with the action and a type of action.