Methods and apparatus for monitoring an audience of media based on thermal imaging
Methods, apparatus, systems, and articles of manufacture are disclosed. An example apparatus includes a thermal image detector to determine a heat blob count based on a frame of thermal image data, the frame of thermal image data captured in the media environment, a comparator to compare the heat blob count to a prompted people count, the prompted people count based on one or more responses to a prompting message, and when the heat blob count and the prompted people count match, cause a timer that is to trigger generation of the prompting message to be reset.
1. An apparatus to count people in a media environment, the apparatus comprising:
a thermal image detector to determine a heat blob count based on a frame of thermal image data, the frame of thermal image data corresponding to the media environment;
a comparator to:
compare the heat blob count to a prompted people count, the prompted people count based on one or more responses to a prompting message;
when the heat blob count and the prompted people count match, cause a timer that is to trigger generation of the prompting message to be reset; and
a controller to train a machine learning model based on a feature vector including a verified people count, at least one of a day or time corresponding to the verified people count, and a media source detected in the media environment during the at least one of the day or time, the verified people count based on the comparison of the heat blob count to the prompted people count, the machine learning model to predict subsequent people counts based on features representative of the media environment.
2. The apparatus of claim 1 , wherein the comparator is to cause the prompting message to be generated when the prompted people count does not match the heat blob count.
3. The apparatus of claim 1 , wherein when the comparator determines the prompted people count is greater than the heat blob count, the comparator is to decrease the prompted people count to equal the heat blob count.
4. The apparatus of claim 1 , wherein the controller is a first controller, and further including a second controller to determine the prompted people count based on the one or more responses to the prompting message, the one or more responses associated with unique people identifiers.
5. The apparatus of claim 4 , wherein the second controller is to generate the prompting message to request a response to provide an input or remove an input.
6. The apparatus of claim 1 , wherein the thermal image detector is to evaluate features in the frame of thermal image data to detect one or more human size heat blobs, the one or more human size heat blobs corresponding to the heat blob count.
7. The apparatus of claim 1 , wherein the comparator is to verify a number of people in the media environment based on comparison of the heat blob count and the prompted people count.
8. A method to count people in a media environment, the method comprising:
determining a heat blob count based on a frame of thermal image data, the frame of thermal image data captured in the media environment;
comparing the heat blob count to a prompted people count, the prompted people count based on one or more responses to a prompting message;
when the heat blob count and the prompted people count match, causing a timer that is to trigger generation of the prompting message to be reset; and
training a machine learning model based on a feature vector including a verified people count, at least one of a day or time corresponding to the verified people count, and a media source detected in the media environment during the at least one of the day or time, the verified people count based on the comparison of the heat blob count to the prompted people count, the machine learning model to predict subsequent people counts based on features representative of the media environment.
9. The method of claim 8 , wherein further including causing the prompting message to be generated when the prompted people count does not match the heat blob count.
10. The method of claim 8 , further including decreasing the prompted people count to equal the heat blob count when the prompted people count is greater than the heat blob count.
11. The method of claim 8 , further including determining the prompted people count based on the one or more responses to the prompting message, the one or more responses associated with unique people identifiers.
12. The method of claim 11 , further including generating the prompting message to request a response to provide an input or remove an input.
13. The method of claim 8 , further including evaluating features in the frame of thermal image data to detect one or more human size heat blobs, the one or more human size heat blobs corresponding to the heat blob count.
14. The method of claim 8 , further including verifying a number of people in the media environment based on comparison of the heat blob count and the prompted people count.
15. A non-transitory computer readable storage medium comprising instructions that, when executed, cause one or more processors to at least:
determine a heat blob count based on a frame of thermal image data, the frame of thermal image data captured in a media environment;
compare the heat blob count to a prompted people count, the prompted people count based on one or more responses to a prompting message;
when the heat blob count and the prompted people count match, cause a timer that is to trigger generation of the prompting message to be reset; and
train a machine learning model based on a feature vector including a verified people count, at least one of a day or time corresponding to the verified people count, and a media source detected in the media environment during the at least one of the day or time, the verified people count based on the comparison of the heat blob count to the prompted people count, the machine learning model to predict subsequent people counts based on features representative of the media environment.
16. The non-transitory computer readable storage medium of claim 15 , wherein the instructions, when executed, cause the one or more processors to cause the prompting message to be generated when the prompted people count does not match the heat blob count.
17. The non-transitory computer readable storage medium of claim 15 , wherein the instructions, when executed, cause the one or more processors to decrease the prompted people count to equal the heat blob count when the prompted people count is greater than the heat blob count.
18. The non-transitory computer readable storage medium of claim 15 , wherein the instructions, when executed, cause the one or more processors to determine the prompted people count based on the one or more responses to the prompting message, the one or more responses associated with unique people identifiers.
19. The non-transitory computer readable storage medium of claim 18 , wherein the instructions, when executed, cause the one or more processors to generate the prompting message to request a response to provide an input or remove an input.
20. The non-transitory computer readable storage medium of claim 15 , wherein the instructions, when executed, cause the one or more processors to evaluate features in the frame of thermal image data to detect one or more human size heat blobs, the one or more human size heat blobs corresponding to the heat blob count.
21. The non-transitory computer readable storage medium of claim 15 , wherein the instructions, when executed, cause the one or more processors to verify a number of people in the media environment based on comparison of the heat blob count and the prompted people count.