Methods and apparatus to determine tags for media using multiple media features
Methods, apparatus, systems and articles of manufacture are disclosed to determine tags for unknown media using multiple media features. The methods, apparatus, systems and articles of manufacture for tagging unknown media, extracts audio features from an audio portion of the unknown media. The methods, apparatus, systems and articles of manufacture for tagging unknown media, extracts image features from image portions of the unknown media. The methods, apparatus, systems and articles of manufacture for tagging unknown media, weights the audio features with respect to the image features or weights the image features with respect to the audio features based at least partially on the recognition technology used to extract the feature. The methods, apparatus, systems and articles of manufacture for tagging unknown media, searches a database of pre-tagged media using the weighted features to generate a list of suggested tags for the unknown media. The methods, apparatus, systems and articles of manufacture for tagging unknown media, tags the unknown media with approved tags from the list of suggested tags.
1. A method for tagging unknown media, the method comprising:
extracting audio features from an audio portion of the unknown media;
extracting image features from image portions of the unknown media;
weighting, by executing an instruction with a processor, the audio features and the image features based at least partially on respective recognition technologies used to extract the features, the weighting including assigning a first weight to a first feature extracted by at least one of a logo recognition engine or an optical character recognition (OCR) engine, the weighting including assigning a second weight different from the first weight to a second feature extracted by a speech recognition engine;
searching, by executing an instruction with the processor, a database of pre-tagged media using a combination of the weighted audio features and the weighted image features to generate a list of suggested tags for the unknown media, wherein when a first feature included in the combination of the weighted audio features and the weighted image features is assigned to a first category of a plurality of categories, the searching in the database is limited to pre-tagged media having category values matching the first category of the first feature; and
assigning, by executing an instruction with the processor, tags from the list of suggested tags to the unknown media.
2. The method of claim 1 , wherein the audio features are extracted from the audio portion of the unknown media using at least one of the speech recognition engine, a sound identifier, and a song or tune identifier.
3. The method of claim 1 , wherein the image features are extracted from the image portions of the unknown media using at least one of the logo recognition engine, the optical character recognition (OCR) engine, an object recognition engine and a graphic recognition engine.
4. The method of claim 1 , wherein the list of suggested tags includes relevancy scores indicating how relevant the tags are with respect to the unknown media.
5. The method of claim 4 , wherein a tag in the list of suggested tags is automatically approved when the relevancy score for the tag is above a threshold relevancy score.
6. The method of claim 1 , wherein the first category is one of the following categories: a company, a brand, a product type, a product manufacturer.
7. The method of claim 1 , wherein multiple search engines are used to search the database of pre-tagged media.
8. A method for tagging unknown media, the method comprising:
extracting audio features from an audio portion of the unknown media;
extracting image features from image portions of the unknown media;
weighting, by executing an instruction with a processor, the audio features with respect to the image features and/or the image features with respect to the audio features, the weighting based at least partially on respective recognition technologies used to extract the features, wherein when weighting the features, a highest weight is given to features extracted by a logo recognition engine, a next highest weight is given to features extracted by an optical character recognition (OCR) engine, a next highest weight is given to features extracted by a speech recognition engine and a lowest weight is given to features extracted by a graphic recognition engine;
searching, by executing an instruction with the processor, a database of pre-tagged media using the weighted features to generate a list of suggested tags for the unknown media; and
assigning, by executing an instruction with the processor, tags from the list of suggested tags to the unknown media.
9. The method of claim 8 , wherein the weights used for the logo recognition engine, the OCR engine, the speech recognition engine and the graphic recognition engine are: 10, 5, 2 and 1 respectively.
10. An apparatus to tag unknown media, the apparatus comprising:
a media separator to separate an audio portion from the unknown media and image portions from the unknown media;
an audio extractor to extract audio features from the audio portion of the unknown media;
an image extractor to extract image features from the image portions of the unknown media;
a feature weighter to weight the audio features and the image features based at least partially on respective recognition technologies used to extract the features, the feature weighter to assign a first weight to a first feature extracted by at least one of a logo recognition engine or an optical character recognition (OCR) engine, and to assign a second weight different from the first weight to a second feature extracted by a speech recognition engine;
one or more database search engines to search a database of pre-tagged media using a combination of the weighted audio features and the weighted image features to generate a list of suggested tags for the unknown media, wherein when a first feature included in the combination of the weighted audio features and the weighted image features is assigned to a first category of a plurality of categories, the search in the database is limited to pre-tagged media having category values matching the first category of the first feature; and
a tag assigner to assign tags from the list of suggested tags to the unknown media.
11. The apparatus of claim 10 , wherein the audio features are extracted from the audio portion of the unknown media using at least one of the speech recognition engine, a sound identifier and a song or tune identifier.
12. The apparatus of claim 10 , wherein the image features are extracted from the image portions of the unknown media using at least one of the logo recognition engine, the OCR engine, an object recognition engine and a graphic recognition engine.
13. The apparatus of claim 10 , wherein the list of suggested tags includes relevancy scores indicating how relevant the tags are with respect to the unknown media.
14. The apparatus of claim 13 , wherein a tag in the list of suggested tags is automatically approved when the relevancy score for the tag is above a threshold relevancy score.
15. The apparatus of claim 10 , wherein the first category is one of the following categories: a company, a brand, a product type, a product manufacturer.
16. The apparatus of claim 10 , wherein multiple search engines are used to search the database of pre-tagged media.
17. An apparatus to tag unknown media, the apparatus comprising:
a media separator to separate an audio portion from the unknown media and image portions from the unknown media;
an audio extractor to extract audio features from the audio portion of the unknown media;
an image extractor to extract image features from the image portions of the unknown media;
a feature weighter to weight the audio features with respect to the image features, and/or to weight the image features with respect to the audio features, based at least partially on respective recognition technologies used to extract the features, wherein the feature weighter is to assign a highest weight to features extracted by a logo recognition engine, a next highest weight to features extracted by an optical character recognition (OCR) engine, a next highest weight to features extracted by a speech recognition engine and a lowest weight to features extracted by a graphic recognition engine;
one or more database search engines to search a database of pre-tagged media using the weighted features to generate a list of suggested tags for the unknown media;
a tag assigner to assign tags from the list of suggested tags to the unknown media.
18. The apparatus of claim 17 , wherein the weights used for the logo recognition engine, the OCR engine, the speech recognition engine and the graphic recognition engine are: 10, 5, 2 and 1 respectively.
19. A tangible computer readable medium comprising computer readable instructions which, when executed, cause a processor to at least:
extract audio features from an audio portion of the unknown media;
extract image features from image portions of the unknown media;
weight the audio features and the image features based at least partially on respective recognition technologies used to extract the features, wherein to weight the features, the instructions cause the processor to assign a first weight to a first feature extracted by at least one of a logo recognition engine or an optical character recognition (OCR) engine, and the instructions cause the processor to assign a second weight different from the first weight is given to a second feature extracted by a speech recognition engine;
search a database of pre-tagged media using a combination of the weighted audio features and the weighted image features to generate a list of suggested tags for the unknown media, wherein when a first feature included in the combination of the weighted audio features and the weighted image features is assigned to a first category of a plurality of categories, the search in the database is limited to pre-tagged media having category values matching the first category of the first feature; and
assign tags from the list of suggested tags to the unknown media.
20. The computer readable medium of claim 19 , wherein the audio features are extracted from the audio portion of the unknown media using at least one of the speech recognition engine, a sound identifier and a song or tune identifier.
21. The computer readable medium of claim 19 , wherein the image features are extracted from the image portions of the unknown media using at least one of the logo recognition engine, the OCR engine, an object recognition engine and a graphic recognition engine.
22. The computer readable medium of claim 19 , wherein the list of suggested tags includes relevancy scores indicating how relevant the tags are with respect to the unknown media.
23. The computer readable medium of claim 22 , wherein a tag in the list of suggested tags is automatically approved when a relevancy score for the tag is above a threshold relevancy score.
24. The computer readable medium of claim 19 , wherein the first category is one of the following categories: a company, a brand, a product type, a product manufacturer.
25. The computer readable medium of claim 19 , wherein the instructions cause the processor to use multiple search engines to search the database of pre-tagged media.
26. A tangible computer readable medium comprising computer readable instructions which, when executed, cause a processor to at least:
extract audio features from an audio portion of the unknown media;
extract image features from image portions of the unknown media;
weight the audio features with respect to the image features, and/or weight the image features with respect to the audio features, based at least partially on respective recognition technologies used to extract the features, wherein to weight the features, the instructions cause the processor to assign a highest weight to features extracted by a logo recognition engine, a next highest weight to features extracted by an optical character recognition (OCR) engine, a next highest weight to features extracted by a speech recognition engine and a lowest weight to features extracted by a graphic recognition engine;
search a database of pre-tagged media using the weighted features to generate a list of suggested tags for the unknown media;
assign tags from the list of suggested tags to the unknown media.
27. The computer readable medium of claim 26 , wherein the weights used for the logo recognition engine, the OCR engine, the speech recognition engine and the graphic recognition engine are: 10, 5, 2 and 1 respectively.