Methods and apparatus to detect boring media

1. An apparatus to detect boring media, the apparatus comprising: a feature calculator implemented by hardware or at least one processor to determine at least a first feature and a second feature associated with at least one of a video feed or an audio feed of monitored media when a media monitoring code is not detected in the monitored media, the first and second features representing different characteristics of at least one of the video feed or the audio feed; and a boring media event determiner implemented by hardware or the at least one processor to: process the first and second features with a neural network; propagate a code absence alert to a first server when an output of the neural network does not satisfy a threshold; and propagate a boring media event notification to a second server when the output of the neural network does satisfy the threshold.

2. The apparatus of claim 1 , wherein the first and second features correspond to respective different ones of at least one of an image temporal variation value, an image chromatic information value, a sound temporal domain flatness value, or a sound harmonic richness value, and the feature calculator further includes: an image temporal variation calculator to calculate the image temporal variation value based on at least one of a red plane, a green plane, or a blue plane of the video feed of the monitored media; an image chromatic information calculator to calculate the image chromatic information value based on a histogram of at least one of the red plane, the green plane, or the blue plane of the video feed of the monitored media; a sound temporal domain flatness calculator to calculate the sound temporal domain flatness value based on an envelope of the audio feed of the monitored media; and a sound harmonic richness calculator to calculate the sound harmonic richness value based on a Fourier transform of the audio feed of the monitored media.

3. The apparatus of claim 1 , further including a machine learning module to train the neural network based on at least respective first features or second features for a first group of media assets and corresponding boring media detection results for the first group of media assets, the machine learning module including: a boring media detection engine validator to determine a validation output of the trained neural network based on at least respective first features or second features for a second group of media assets and corresponding boring media detection results for the second group of media assets; and a boring media detection engine to deploy the trained neural network to the boring media event determiner when the validation output of the trained neural network satisfies the threshold.

4. The apparatus of claim 3 , wherein the boring media detection engine validator instructs the neural network to retrain when the validation output of the neural network does not satisfy the threshold.

5. The apparatus of claim 3 , wherein to determine the validation output of the trained neural network, the boring media detection engine validator is to: execute the trained neural network based on at least respective first features or second features for the second group of media assets with the boring media detection engine; and compare trained neural network outputs with the corresponding boring media detection results for the second group of media assets.

6. The apparatus of claim 3 , wherein the machine learning module updates the trained neural network with at least one of the first feature or the second feature associated with the monitored media and corresponding boring media detection result for the monitored media.

7. The apparatus of claim 3 , wherein the feature calculator calculates a third feature associated with at least one of the video feed or the audio feed of the monitored media, the machine learning module to update the trained neural network based on at least one of the first, second, or third features.

8. A non-transitory computer readable storage medium for detecting boring media comprising instructions that, when executed, cause a machine to at least: determine at least a first feature and a second feature associated with at least one of a video feed or an audio feed of monitored media when a media monitoring code is not detected in the monitored media, the first and second features representing different characteristics of at least one of the video feed or the audio feed; process the first and second features with a neural network; propagate a code absence alert to a first server when an output of the neural network does not satisfy a threshold; and propagate a boring media event notification to a second server when the output of the neural network does satisfy the threshold.

9. The non-transitory computer readable storage medium of claim 8 , wherein the first and second features correspond to respective different ones of at least one of an image temporal variation value, an image chromatic information value, a sound temporal domain flatness value, or a sound harmonic richness value and the instructions to calculate at least one of the first feature and the second feature further cause the machine to at least: calculate the image temporal variation value based on at least one of a red plane, a green plane, or a blue plane of the video feed of the monitored media; calculate the image chromatic information value based on a histogram of at least one of the red plane, the green plane, or the blue plane of the video feed of the monitored media; calculate the sound temporal domain flatness value based on an envelope of the audio feed of the monitored media; and calculate the sound harmonic richness value based on a Fourier transform of the audio feed of the monitored media.

10. The non-transitory computer readable storage medium of claim 8 , wherein the instructions, when executed to generate the neural network, cause the machine to at least: train the neural network based on at least respective first features or second features for a first group of media assets and corresponding boring media detection results for the first group of media assets; determine a validation output for the neural network based on at least respective first features or second features for a second group of media assets and corresponding boring media detection results for the second group of media assets to; and deploy the trained neural network when the validation output of the neural network satisfies a threshold.

11. The non-transitory computer readable storage medium of claim 10 , wherein the instructions, when executed, further cause the machine to at least retrain the neural network when the validation output of the trained neural network does not satisfy the threshold.

12. The non-transitory computer readable storage medium of claim 10 , wherein the instructions, when executed to determine the validation output of the neural network, further cause the machine to at least: execute the trained neural network based on respective first features or second features for the second group of media assets; and compare the trained neural network outputs with the corresponding boring media detection results for the second group of media assets.

13. The non-transitory computer readable storage medium of claim 10 , wherein the instructions, when executed, further cause the machine to update the trained neural network with the first feature or the second feature associated with the monitored media and corresponding boring media detection result for the monitored media.

14. The non-transitory computer readable storage medium of claim 10 , wherein the instructions are further to cause the machine to at least: determine a third feature associated with at least one of the video feed or the audio feed of the monitored media; and update the trained neural network based on at least one of the first, second, or third features.

15. A method for detecting boring media, the method comprising:

detecting a media monitoring code is absent in monitored media;

determining at least a first feature and a second feature associated with at least one of a video feed or an audio feed of the monitored media, the first and second features representing different characteristics of at least one of the video feed or the audio feed;

processing the first and second features with a neural network;

in response to an output of the neural network not satisfying a threshold, propagating a code absence alert to a first server; and

in response the output of the neural network satisfying the threshold, propagating a boring media event notification to a second server.

16. The method of claim 15 , wherein the first and second features correspond to respective different ones of at least one of an image temporal variation value, an image chromatic information value, a sound temporal domain flatness value, or a sound harmonic richness value, and further including:

calculating the image temporal variation value based on at least one of a red plane, a green plane, or a blue plane of the video feed of the monitored media;

calculating the image chromatic information value based on a histogram of at least one of the red plane, the green plane, or the blue plane of the video feed of the monitored media;

calculating the sound temporal domain flatness value based on an envelope of the audio feed of the monitored media; and calculating the sound harmonic richness value based on a Fourier transform of the audio feed of the monitored media.

17. The method of claim 15 , wherein the neural network is generated by:

training the neural network based on respective first features or second features for a first group of media assets and corresponding boring media detection results for the first group of media assets;

determining a validation output of the trained neural network based on respective first features or second features for a second group of media assets and corresponding boring media detection results for the second group of media assets; and

in response to the validation output of the trained neural network satisfying a threshold, deploying the trained neural network.

18. The method of claim 17 , further including, in response to the validation output of the trained neural network not satisfying the threshold, retraining the neural network.

19. The method of claim 17 , wherein determining the validation output of the neural network further includes: executing the neural network based on respective first features or second features for the second group of media assets; and comparing neural network outputs with the corresponding boring media detection results for the second group of media assets.

20. The method of claim 17 , further including: determining a third feature associated with at least one of the video feed or the audio feed of the monitored media; and updating the trained neural network based on at least one of the first, second, or third features.