Viewer reaction detection and content control
Systems, apparatuses, and methods for detecting the reactions of primary and secondary viewers of content are described. Reactions of a primary or secondary viewer of content may be detected through use of a sensor and machine learning model. Based on the reaction of the primary or secondary viewer satisfying some criteria, outputting of the content may be modified and/or alternative content may be provided. Furthermore, metadata may be generated based on the detection of adverse reactions to intrusions by a viewer of content that is associated with an indication that the outputted content is associated with certain predefined types.
1. A method comprising:
outputting, by a computing device, content;
detecting, during the outputting of the content, based on data received from one or more sensors, a presence of a primary viewer of the content being outputted;
determining a reaction of the primary viewer to one or more secondary viewers, wherein the secondary viewers were not present during the detecting the presence of the primary viewer;
determining, during the outputting of the content, using a machine learning model, and based on the reaction of the primary viewer, a reaction of the one or more secondary viewers, different from the primary viewer, to a portion of the content being outputted; and
modifying, based on the determined reaction of the one or more secondary viewers satisfying one or more criteria, the outputting of the content.
2. The method of claim 1 , wherein the detecting the presence of the primary viewer comprises detecting the presence of the primary viewer using the machine learning model.
3. The method of claim 1 , wherein the modifying the outputting of the content comprises one or more of:
stopping the outputting of audio of the content;
reducing a volume of audio of the content;
stopping the outputting of video of the content; or
outputting alternative content that is different from the content.
4. The method of claim 1 , wherein the determined reaction of the one or more secondary viewers comprises one or more of:
the one or more secondary viewers viewing the content,
the one or more secondary viewers looking at the primary viewer, or
one or more of a surprised facial expression, an embarrassed facial expression, or a disgusted facial expression.
5. The method of claim 1 , further comprising:
determining a first portion of the content associated with one or more predefined types of content, and wherein the modifying the outputting of the content comprises:
outputting, instead of the first portion, a second portion of the content.
6. The method of claim 1 , wherein the content is associated with metadata that indicates one or more portions of content that are associated with one or more predefined types of content, and the determined reaction of the one or more secondary viewers comprises a reaction at a time that a portion of content, associated with the metadata, is being outputted.
7. The method of claim 1 , wherein the machine learning model comprises a Bayesian hierarchical model and is configured to determine the reaction of the one or more secondary viewers.
8. The method of claim 1 , wherein the machine learning model is configured to determine a reaction based on an amplitude of an utterance of the one or more secondary viewers.
9. The method of claim 1 , wherein the machine learning model is configured to determine a reaction based on recognition of words spoken by the one or more secondary viewers.
10. The method of claim 1 , wherein the detecting the presence of the primary viewer comprises:
determining that a viewer that initiated the outputting of the content is the primary viewer.
11. The method of claim 1 , wherein one or more portions of the content are associated with metadata comprising one or more key words, and wherein the determined reaction of the one or more secondary viewers comprises a reaction determined to occur at a time that a portion of the content associated with the one or more key words is being outputted.
12. The method of claim 1 , wherein the modifying the outputting of the content comprises stopping the outputting of at least a portion of the content and outputting alternative content comprising an advertisement.
13. The method of claim 1 , wherein the modifying the outputting comprises outputting alternative content that comprises an advertisement, the method further comprising:
determining, using a machine learning model, one or more estimated ages of the one or more secondary viewers; and
determining the advertisement based on a lowest of the one or more estimated ages.
14. The method of claim 1 , further comprising:
detecting, during outputting of the content to the primary viewer, an intrusion into an environment associated with the primary viewer;
determining a reaction of the primary viewer to the intrusion; and
based on the reaction of the primary viewer to the intrusion, and based on determining that metadata associated with the content comprises one or more key words associated with one or more predefined types of content, generating additional metadata indicating that the content is provocative and a time associated with the reaction of the primary viewer to the intrusion.
15. A method comprising:
detecting, during outputting of content to a primary viewer and based on first data received from one or more sensors, a first intrusion into an environment associated with the primary viewer;
determining, based on second data received from one or more sensors, a reaction of the primary viewer to the first intrusion; and
in response to the first intrusion, based on the reaction of the primary viewer, and based on determining that metadata associated with the content comprises one or more key words associated with one or more predefined types of content, modifying the outputting of the content.
16. The method of claim 15 , wherein the metadata comprises one or more of:
a transcript corresponding to speech in the content or
a description of one or more non-speech portions of the content.
17. The method of claim 15 , wherein the first intrusion comprises one or more of:
another person coming within a predefined distance of the primary viewer,
a door opening, or
a light being turned on or off.
18. A method comprising:
determining, during outputting of content, based on data received from one or more sensors, a reaction of a primary viewer that indicates a first presence of one or more secondary viewers of the content; and
in response to the first presence, modifying, based on the determined reaction satisfying one or more criteria, the outputting of the content.
19. The method of claim 18 , wherein the one or more sensors comprise a heart-rate sensor configured to detect a heart-rate of the primary viewer, and wherein the reaction of the primary viewer that indicates the first presence of one or more secondary viewers is based on fluctuations of the heart-rate of the primary viewer.
20. The method of claim 18 , wherein the one or more sensors comprise a camera, and wherein the reaction of the primary viewer that indicates the first presence of one or more secondary viewers comprises the primary viewer gazing away from the content being outputted for greater than a threshold amount of time.